<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:googleplay="http://www.google.com/schemas/play-podcasts/1.0"><channel><title><![CDATA[Listen to Decide]]></title><description><![CDATA[How customer conversations lead to better decisions. A newsletter by ReveLumi.]]></description><link>https://blog.revelumi.com</link><image><url>https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png</url><title>Listen to Decide</title><link>https://blog.revelumi.com</link></image><generator>Substack</generator><lastBuildDate>Mon, 31 Aug 2026 15:00:34 GMT</lastBuildDate><atom:link href="https://blog.revelumi.com/feed" rel="self" type="application/rss+xml"/><copyright><![CDATA[Joca Torres]]></copyright><language><![CDATA[en]]></language><webMaster><![CDATA[listentodecide@substack.com]]></webMaster><itunes:owner><itunes:email><![CDATA[listentodecide@substack.com]]></itunes:email><itunes:name><![CDATA[Joca Torres]]></itunes:name></itunes:owner><itunes:author><![CDATA[Joca Torres]]></itunes:author><googleplay:owner><![CDATA[listentodecide@substack.com]]></googleplay:owner><googleplay:email><![CDATA[listentodecide@substack.com]]></googleplay:email><googleplay:author><![CDATA[Joca Torres]]></googleplay:author><itunes:block><![CDATA[Yes]]></itunes:block><item><title><![CDATA[Agents need a harness]]></title><description><![CDATA[How we built the ReveLumi agent to listen, understand, and question respondents to extract useful insights for our customers.]]></description><link>https://blog.revelumi.com/p/agents-need-a-harness</link><guid isPermaLink="false">https://blog.revelumi.com/p/agents-need-a-harness</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 24 Aug 2026 15:57:22 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/agentes-precisam-de-redeas">Portuguese version</a>]</p><p>To build an agent that can listen, understand, and question customers to extract useful insights, conversational capability from LLMs isn't enough on its own. We need a harness.</p><p>A harness is what turns a horse&#8217;s raw power into useful movement. Without it, the horse runs, but it doesn&#8217;t go where it needs to go. The same logic applies to AI. Models are extremely powerful, but they need a harness to direct that power toward the right place and the results we&#8217;re after.</p><h2>The agency mental model</h2><p>When we present <a href="https://ReveLumi.com">ReveLumi</a> to prospective customers, we like to use the mental model of hiring a research agency. When we hire an agency, the first step is handing over a briefing.</p><p>That briefing carries context about the company and the problem, the research objectives, the hypotheses that need to be validated, and the decisions that depend on the outcome. It&#8217;s what gives the agent a reason to ask the most appropriate questions to reach the research objectives.</p><p>Beyond the general briefing, each respondent has an individual context field. The agent doesn&#8217;t treat every respondent as the same generic person. It walks into the conversation already knowing who&#8217;s on the other side, which changes the kind of questions it asks, how it interprets the answers, and how it steers the conversation.</p><h2>Memory as a guardrail</h2><p>Every generative AI has a temptation: when it doesn&#8217;t know something, it makes up a plausible answer. In a research interview, that&#8217;s the opposite of what we want. A hallucination becomes a false data point, and a false data point becomes a wrong decision.</p><p>That&#8217;s why we built a guardrail: when the agent doesn&#8217;t know the answer to something the respondent asks, it says it doesn&#8217;t have that information right now, that it will check, and keeps the interview going without freezing or making things up. The conversation continues, trust isn&#8217;t lost, and what gets recorded is real.</p><h2>What the harness delivers</h2><p>None of this shows up to the person being interviewed as &#8220;AI rules.&#8221; It shows up as a conversation that makes sense, that remembers what was already said, that knows when it doesn&#8217;t know. And to the person on the other side, receiving the final report, it shows up as something valuable: research that actually answers what needed to be answered.</p><p>The difference between using a generic AI to talk to customers and using an agent that listens to generate decisions isn&#8217;t in the model&#8217;s intelligence. It&#8217;s in the harness we use to guide the AI toward the results we need.</p><p>Want to see how this harness works in a real interview? <a href="https://share.revelumi.com/c/06VP6MA2?lang=en">Talk to our agent</a> by answering our demo survey about how you choose what to watch on Netflix.</p><p>And if you&#8217;d like to put this agent to work researching with your customers, <a href="https://calendar.app.google/b5WAKVv46cD3zrQb6">book a demo</a>! We&#8217;d love to show you our agent and its harness.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div>]]></content:encoded></item><item><title><![CDATA[Agentes precisam de rédeas]]></title><description><![CDATA[Como constru&#237;mos a agente da ReveLumi para ouvir, entender e questionar respondentes para extrair insights &#250;teis para nossas clientes.]]></description><link>https://blog.revelumi.com/p/agentes-precisam-de-redeas</link><guid isPermaLink="false">https://blog.revelumi.com/p/agentes-precisam-de-redeas</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 24 Aug 2026 15:52:27 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Para construir uma agente capaz de ouvir, entender e questionar clientes para extrair insights &#250;teis, n&#227;o basta contar apenas com a capacidade conversacional das LLMs. Precisamos de r&#233;deas ou, em ingl&#234;s, de <em>harness</em>. </p><p>A r&#233;dea &#233; a pe&#231;a do arreio que transforma a for&#231;a bruta de um cavalo em movimento &#250;til. Sem ela, o cavalo corre, mas n&#227;o vai aonde precisa ir. Com IA &#233; a mesma l&#243;gica. Os modelos s&#227;o muito poderosos, mas precisam de r&#233;deas para serem direcionados ao lugar certo e aos resultados que buscamos.</p><h2>O modelo mental da ag&#234;ncia</h2><p>Ao apresentarmos a ReveLumi, gostamos de usar o modelo mental da contrata&#231;&#227;o de uma ag&#234;ncia de pesquisa. Quando contratamos uma ag&#234;ncia, o primeiro passo &#233; apresentar um briefing.</p><p>Esse briefing apresentar&#225; o contexto da empresa e do problema, os objetivos da pesquisa, as hip&#243;teses a serem validadas e as decis&#245;es que dependem do resultado. &#201; o que d&#225; &#224; agente um motivo para fazer as perguntas mais apropriadas para conseguir atingir os objetivos da pesquisa.</p><p>Al&#233;m do briefing geral, cada pessoa respondente tem um campo de contexto individual. A agente n&#227;o trata todas as respondentes como uma pessoa gen&#233;rica. Ela chega &#224; conversa j&#225; sabendo quem est&#225; do outro lado, o que muda o tipo de pergunta que faz, a forma como interpreta a resposta e a condu&#231;&#227;o da conversa.</p><h2>A mem&#243;ria como guardrail</h2><p>Toda IA generativa tem uma tenta&#231;&#227;o: quando n&#227;o sabe algo, inventa uma resposta plaus&#237;vel. Numa entrevista de pesquisa, isso &#233; o oposto do que buscamos. Uma alucina&#231;&#227;o vira um dado falso, e um dado falso vira uma decis&#227;o errada.</p><p>Por isso, constru&#237;mos um guardrail: quando a agente n&#227;o sabe a resposta a algo que a pessoa entrevistada pergunta, ela diz que n&#227;o tem essa informa&#231;&#227;o agora, que vai checar e segue com a entrevista sem travar nem inventar. A conversa continua, a confian&#231;a n&#227;o se perde, e o que fica registrado &#233; real.</p><h2>O que a r&#233;dea entrega</h2><p>Nada disso aparece para quem est&#225; sendo entrevistado como &#8220;regras de IA&#8221;. Aparece como uma conversa que faz sentido, que lembra o que j&#225; foi dito, que sabe quando n&#227;o sabe. E para quem est&#225; do outro lado, recebendo o relat&#243;rio final, aparece como algo muito valioso: uma pesquisa que efetivamente responde ao que precisava ser respondido.</p><p>A diferen&#231;a entre usar uma IA gen&#233;rica para conversar com clientes e usar uma agente que ouve para gerar decis&#245;es n&#227;o est&#225; na intelig&#234;ncia do modelo. Est&#225; nas r&#233;deas que utilizamos para guiar a IA para gerar os resultados que precisamos.</p><p>Quer ver como essa r&#233;dea funciona numa entrevista de verdade? <a href="https://share.revelumi.com/c/06VP6MA2">Converse com nossa agente</a> respondendo &#224; nossa pesquisa demo sobre como voc&#234; escolhe o que assistir na Netflix.</p><p>E se voc&#234; quiser colocar essa agente para fazer pesquisas com suas clientes, <a href="https://calendar.app.google/b5WAKVv46cD3zrQb6">marque uma demo</a>! Ser&#225; um prazer mostrar nossa agente e suas r&#233;deas.</p><p></p>]]></content:encoded></item><item><title><![CDATA[Cansado de NPS que não te diz o que fazer?]]></title><description><![CDATA[A gente tamb&#233;m estava. Por isso criamos o NPS Conversacional.]]></description><link>https://blog.revelumi.com/p/cansado-de-nps-que-nao-te-diz-o-que</link><guid isPermaLink="false">https://blog.revelumi.com/p/cansado-de-nps-que-nao-te-diz-o-que</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 17 Aug 2026 11:25:15 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!Ftr4!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/tired-of-an-nps-that-doesnt-tell">English version</a>]</p><p>O NPS virou padr&#227;o porque &#233; simples: uma pergunta, uma nota, um n&#250;mero. Mas essa simplicidade tem um custo que a maioria das empresas sente, mas poucas nomeiam.</p><h3>O problema n&#227;o &#233; a nota, mas o que ela esconde.</h3><p>Uma pessoa que d&#225; nota 0 e outra que d&#225; nota 6 caem na mesma categoria: detratoras. Uma nota 7 e uma nota 8 tamb&#233;m s&#227;o tratadas como id&#234;nticas: neutras. Duas experi&#234;ncias completamente diferentes, reduzidas ao mesmo r&#243;tulo.</p><p>Some isso ao campo de coment&#225;rio, que &#233; opcional. A maioria n&#227;o escreve nada. Quem escreve, geralmente escreve uma linha. &#8220;Bom.&#8221; &#8220;Poderia melhorar.&#8221; Voc&#234; fica com um n&#250;mero que traz pouca informa&#231;&#227;o e um coment&#225;rio raso demais para preencher essa lacuna.</p><p>O resultado: voc&#234; sabe se est&#225; indo bem ou mal, mas raramente sabe o porqu&#234; e, consequentemente, o que fazer a respeito. &#201; a cr&#237;tica mais repetida sobre o NPS e &#233; justa. O n&#250;mero n&#227;o &#233; acion&#225;vel sozinho.</p><h3>A gente tamb&#233;m sentia isso</h3><p>Desde minha &#233;poca na Locaweb, realizamos pesquisas de NPS e sempre sofri com essa falta de informa&#231;&#227;o acion&#225;vel. Isso sempre me frustrou. Anderson e eu rod&#225;vamos NPS na Conta Azul, com a mesma frustra&#231;&#227;o.</p><p>Por isso criamos o NPS Conversacional na ReveLumi. A pergunta continua sendo a mesma: de 0 a 10, o quanto a pessoa recomendaria determinado produto ou servi&#231;o a um amigo? Mas, depois da nota, a agente busca extrair o que h&#225; por tr&#225;s dela. Ela pergunta e pergunta de novo se precisar, at&#233; a motiva&#231;&#227;o real aparecer.</p><p>Para demonstrar, rodamos uma pesquisa de NPS do LinkedIn e veja s&#243; o que achamos.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!Ftr4!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!Ftr4!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 424w, https://substackcdn.com/image/fetch/$s_!Ftr4!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 848w, https://substackcdn.com/image/fetch/$s_!Ftr4!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 1272w, https://substackcdn.com/image/fetch/$s_!Ftr4!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!Ftr4!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png" width="1456" height="819" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:819,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:1089984,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/211540263?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!Ftr4!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 424w, https://substackcdn.com/image/fetch/$s_!Ftr4!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 848w, https://substackcdn.com/image/fetch/$s_!Ftr4!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 1272w, https://substackcdn.com/image/fetch/$s_!Ftr4!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2f597f1d-ae53-4c5e-bb47-8fb81e11d2dc_3200x1800.png 1456w" sizes="100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p><a href="https://share.revelumi.com/c/H08GABDR">Experimente dar sua nota de NPS do LinkedIn</a> e tenha a experi&#234;ncia de ser entrevistado pela nossa agente.</p><p>Depois, fique &#224; vontade para conhecer o <a href="https://share.revelumi.com/r/EKvQ3LbX1TPq0GZCU1rw9NS1SPSA8zmBbJzwsYWlxwk">relat&#243;rio de NPS do LinkedIn</a> repleto de insights acion&#225;veis.</p><h3>Dados dizem o qu&#234;. Clientes dizem o porqu&#234;.</h3><p>O NPS continua sendo &#250;til como term&#244;metro. O que muda &#233; o que vem depois da nota. Em vez de um campo em branco ou uma linha gen&#233;rica, uma conversa que vai at&#233; o fundo do motivo.</p><div><hr></div><p><a href="https://calendar.app.google/HaXstKpqNDuowS3J7">Marque uma demo</a><span> para conhecer mais sobre ReveLumi e o NPS Conversacional.</span></p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[Tired of an NPS that doesn't tell you what to do?]]></title><description><![CDATA[We were too. That's why we built Conversational NPS.]]></description><link>https://blog.revelumi.com/p/tired-of-an-nps-that-doesnt-tell</link><guid isPermaLink="false">https://blog.revelumi.com/p/tired-of-an-nps-that-doesnt-tell</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 17 Aug 2026 11:24:30 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!fmVk!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/cansado-de-nps-que-nao-te-diz-o-que">Portuguese version</a>]</p><p>NPS became the standard because it&#8217;s simple: one question, one score, one number. But that simplicity comes with a cost that most companies feel and few name.</p><h3>The problem isn&#8217;t the score, it&#8217;s what it hides.</h3><p>Someone who gives a 0 and someone who gives a 6 fall into the same category: detractors. A 7 and an 8 are also treated as identical: passives. Two completely different experiences, reduced to the same label.</p><p>Add to that the comment field, which is optional. Most people leave it blank. Those who do write something usually write one line. &#8220;Good.&#8221; &#8220;Could be better.&#8221; You&#8217;re left with a number that already carries little information, and a comment too thin to fill the gap.</p><p>The result: you know whether things are going well or badly, but you rarely know why, and, as a consequence, what to do about it. It&#8217;s the most repeated criticism of NPS, and it&#8217;s a fair one. The number isn&#8217;t actionable on its own.</p><h3>We felt this too</h3><p>Since my time at Locaweb, we&#8217;ve run NPS surveys, and I always struggled with that lack of actionable information. It always frustrated me. Anderson and I ran NPS at Conta Azul with the same frustration.</p><p>That&#8217;s why we built Conversational NPS at ReveLumi. The question stays the same: on a scale of 0 to 10, how likely would you be to recommend this product or service to a friend? But after the score, the agent digs for what&#8217;s behind it. It asks, and asks again if needed, until the real reason comes through.</p><p>To show what that looks like, we ran an NPS survey about LinkedIn. Take a look at what we found.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!fmVk!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!fmVk!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 424w, https://substackcdn.com/image/fetch/$s_!fmVk!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 848w, https://substackcdn.com/image/fetch/$s_!fmVk!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 1272w, https://substackcdn.com/image/fetch/$s_!fmVk!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!fmVk!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png" width="1456" height="819" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/fed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:819,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:1056309,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/211542814?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!fmVk!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 424w, https://substackcdn.com/image/fetch/$s_!fmVk!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 848w, https://substackcdn.com/image/fetch/$s_!fmVk!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 1272w, https://substackcdn.com/image/fetch/$s_!fmVk!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffed88599-6ced-49c2-9478-a3577f18e57a_3200x1800.png 1456w" sizes="100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p><a href="https://share.revelumi.com/c/H08GABDR?lang=en">Try giving your own LinkedIn NPS score</a> and experience being interviewed by our agent yourself.</p><p>Then, feel free to check out the <a href="https://share.revelumi.com/r/EKvQ3LbX1TPq0GZCU1rw9NS1SPSA8zmBbJzwsYWlxwk?lang=en">LinkedIn NPS report</a>, full of actionable insights.</p><p>Data tells us what happened. Customers tell us why.</p><p>NPS is still useful as a thermometer. What changes is what happens after the score. Instead of a blank field or a generic line, a conversation that gets to the bottom of the reason.</p><div><hr></div><p><a href="https://calendar.app.google/HaXstKpqNDuowS3J7">Book a demo</a> to learn more about ReveLumi and Conversational NPS.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[The difference between a comment and a conversation]]></title><description><![CDATA[Often, the comment left after an NPS score isn't enough to understand the motivations behind it.]]></description><link>https://blog.revelumi.com/p/the-difference-between-a-comment</link><guid isPermaLink="false">https://blog.revelumi.com/p/the-difference-between-a-comment</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 10 Aug 2026 11:11:42 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!7g9h!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<h2>The difference between a comment and a conversation</h2><h4>Often, the comment left after an NPS score isn&#8217;t enough to understand the motivations behind it.</h4><p>Every NPS survey includes a comment field after the score. But since that field is optional, many people don&#8217;t answer it, or when they do, they write very little, which doesn&#8217;t give us much to work with.</p><p>That&#8217;s why we built conversational NPS. The agent reads what the person wrote, and when the reason behind the score still isn&#8217;t clear, it asks again.</p><p>A recent case shows this well. Someone gave LinkedIn a score of 10. We asked why, and the answer was:</p><p>&#8220;Actually, my score is based on my experience with LinkedIn Premium.&#8221;</p><p>If this were a regular NPS survey, it would stop here, leaving very little: a high score, a product name, no context on why.</p><p>The agent replied:</p><p>&#8220;Got it, Premium is the big differentiator for you. Can you tell me about a specific moment or something that happened that really showed that value in practice?&#8221;</p><p>And the answer came back complete:</p><p>&#8220;I&#8217;d been unemployed for two months, and in two weeks I landed two job opportunities, plus made contact with a lot of people because I could send unlimited messages to the people I was interested in working with.&#8221;</p><p>Here's the print of the conversation in Portuguese:</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!7g9h!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!7g9h!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 424w, https://substackcdn.com/image/fetch/$s_!7g9h!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 848w, https://substackcdn.com/image/fetch/$s_!7g9h!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 1272w, https://substackcdn.com/image/fetch/$s_!7g9h!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!7g9h!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png" width="1268" height="1052" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/a568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1052,&quot;width&quot;:1268,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:206316,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/210585722?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!7g9h!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 424w, https://substackcdn.com/image/fetch/$s_!7g9h!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 848w, https://substackcdn.com/image/fetch/$s_!7g9h!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 1272w, https://substackcdn.com/image/fetch/$s_!7g9h!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa568d44d-fff8-4d5b-b0db-313ff76b4bed_1268x1052.png 1456w" sizes="100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>It translates into:</p><blockquote><p><em>Wow, a 10! I&#8217;m really thrilled with that. Thinking about your overall experience with LinkedIn, what stood out most for you to give this score? P.S.: Feel free to send an audio message if you prefer.</em></p><p><em>&#8220;Actually, my score is based on my experience with LinkedIn Premium.&#8221;</em></p><p><em>Got it, Premium is the big differentiator for you. Can you tell me about a specific moment or something that happened that really showed that value in practice?</em></p><p><em>&#8220;I&#8217;d been unemployed for two months, and in two weeks I landed two job opportunities, plus made contact with a lot of people because I could send unlimited messages to the people I was interested in working with.&#8221;</em></p></blockquote><p>That wasn&#8217;t in the first comment. It only came up because we asked a second time, asking for an example.</p><p>Asking for the reason makes the person summarize, and a summary turns into a label: Premium, the algorithm, the network. Asking for a moment makes the person remember, and remembering brings the whole scene back, who they were in it, what was at stake, what happened.</p><p>Data tells us what happened. Customers tell us why. And sometimes the why only shows up when we can go deeper, the way we do with <a href="https://revelumi.com/#nps">Conversational NPS</a>.</p><div><hr></div><p><a href="https://calendar.app.google/HaXstKpqNDuowS3J7">Book a demo</a> to learn more about ReveLumi and Conversational NPS.</p>]]></content:encoded></item><item><title><![CDATA[A diferença entre um comentário e uma conversa]]></title><description><![CDATA[Muitas vezes o coment&#225;rio sobre a nota de NPS dada n&#227;o &#233; suficiente para entender as motiva&#231;&#245;es]]></description><link>https://blog.revelumi.com/p/a-diferenca-entre-um-comentario-e</link><guid isPermaLink="false">https://blog.revelumi.com/p/a-diferenca-entre-um-comentario-e</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 10 Aug 2026 11:03:51 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!9nXi!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Toda pesquisa de NPS inclui um campo de coment&#225;rio ap&#243;s a nota. S&#243; que, por esse campo ser opcional, muitas pessoas n&#227;o respondem ou, quando respondem, escrevem pouca coisa, o que n&#227;o nos permite extrair bons insights.</p><p>Por esse motivo, criamos o NPS conversacional. A agente l&#234; o que a pessoa escreveu e, quando a motiva&#231;&#227;o da nota ainda n&#227;o est&#225; clara, pergunta de novo.</p><p>Um caso recente mostra isso bem. Algu&#233;m deu nota 10 ao LinkedIn. Perguntamos o motivo, e a resposta foi:</p><p>&#8220;Na verdade, minha nota est&#225; baseada na minha experi&#234;ncia com LinkedIn Premium.&#8221;</p><p>Se fosse uma pesquisa de NPS comum, a pesquisa pararia aqui, e sobraria pouca coisa: nota alta, um nome de produto, nenhum contexto do porqu&#234;.</p><p>A agente respondeu:</p><p>&#8220;Entendi, o Premium &#233; o grande diferencial para voc&#234;. Consegue me contar sobre um momento espec&#237;fico ou algo que aconteceu que mostrou bem esse valor da assinatura na pr&#225;tica?&#8221;</p><p>E a resposta veio completa:</p><p>&#8220;Fiquei desempregada h&#225; dois meses e em duas semanas consegui duas oportunidades de trabalho, al&#233;m de ter feito contato com muita gente pois podia mandar mensagens ilimitadas para as pessoas que eu tinha interesse de trabalhar.&#8221;</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!9nXi!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!9nXi!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 424w, https://substackcdn.com/image/fetch/$s_!9nXi!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 848w, https://substackcdn.com/image/fetch/$s_!9nXi!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 1272w, https://substackcdn.com/image/fetch/$s_!9nXi!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!9nXi!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png" width="1268" height="1052" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1052,&quot;width&quot;:1268,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:206316,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/210584051?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!9nXi!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 424w, https://substackcdn.com/image/fetch/$s_!9nXi!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 848w, https://substackcdn.com/image/fetch/$s_!9nXi!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 1272w, https://substackcdn.com/image/fetch/$s_!9nXi!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42ccb06b-e316-4a0e-b4c2-1ad804d30902_1268x1052.png 1456w" sizes="100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p></p><p>Isso n&#227;o estava no primeiro coment&#225;rio, s&#243; apareceu porque perguntamos pela segunda vez, pedindo por um exemplo.</p><p>Pedir o motivo faz a pessoa resumir, e resumo vira r&#243;tulo: o Premium, o algoritmo, a rede de contatos. Pedir um momento faz a pessoa lembrar, e lembrar traz a cena inteira de volta, quem ela era ali, o que estava em jogo, o que aconteceu.</p><p>Dados dizem o qu&#234;. Clientes dizem o porqu&#234;. E, &#224;s vezes, o porqu&#234; s&#243; aparece quando conseguimos nos aprofundar como fazemos com o <a href="https://revelumi.com/#nps">NPS Conversacional</a>.</p><div><hr></div><p><a href="https://calendar.app.google/HaXstKpqNDuowS3J7">Marque uma demo</a> para conhecer mais sobre ReveLumi e o NPS Conversacional.</p>]]></content:encoded></item><item><title><![CDATA[ReveLumi competes with the absence of research]]></title><description><![CDATA[The missing habit of talking to customers]]></description><link>https://blog.revelumi.com/p/revelumi-competes-with-the-absence</link><guid isPermaLink="false">https://blog.revelumi.com/p/revelumi-competes-with-the-absence</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 03 Aug 2026 18:57:17 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/revelumi-compete-com-a-ausencia-de">Portuguese version</a>]</p><p>We ran a simple study on how people make decisions on customer-related topics. We got more than 40 responses.</p><p>Most people decide first based on the roadmap, on OKRs, and on whatever leadership has already defined. When information about the customer is missing, the answer is almost never a research study. It&#8217;s booking a 20-minute meeting, sitting next to a colleague, or asking whoever is closest to support. Structured qualitative research barely exists in the day-to-day.</p><p>That changes the question we&#8217;d been asking. It&#8217;s not &#8220;how do we convince someone to switch from the research tool they already use?&#8221;. It&#8217;s &#8220;how do we compete with improvisation and the missing habit of talking to the customer?&#8221;.</p><h2>The Pix that wasn&#8217;t broken</h2><p>One of the richest accounts came from a product owner who handles Pix at a payments institution. Her team surfaced the ten biggest customer complaints, and one kept coming up: &#8220;I can&#8217;t make a Pix transfer, I can&#8217;t even register a key&#8221;.</p><p>The first instinct of any executive, she said, is to assume the app is broken. But when she cross-referenced the tickets with a tool that centralizes support channels, she found something else. Roughly 80% of those customers had a fraud record somewhere, at any bank, not necessarily their own. A Central Bank rule from February this year restricts Pix for anyone with a fraud record on file, and the restriction applies across every institution.</p><p>The data solved the mystery. But the PO herself admits the data isn&#8217;t enough. &#8220;In general, it&#8217;s really hard to understand what the customer is thinking,&#8221; she said. Once the technical problem is explained, the qualitative part is still missing: how the customer reacts to finding this out, what they expect the bank to do, whether the current communication is even clear. The quantitative data answered what. Nobody had answered why.</p><h2>&#8220;They think we&#8217;re a diner&#8221;</h2><p>Not every answer came as a full story. One PM who&#8217;d just joined her company said she asks more senior colleagues when she lacks context about the customer. We asked whether there was any room to understand the customer before deciding, given the delivery pressure she described.</p><p>The answer was one sentence: &#8220;there isn&#8217;t, because they think we&#8217;re a diner&#8221;.</p><p>No further explanation needed. Order taken, delivered, next. It&#8217;s the culture most product companies live in, even the ones that say they put the customer at the center.</p><h2>ChatGPT doesn&#8217;t replace the customer</h2><p>A solo founder described another common shortcut. With no neutral customer to consult, she talks to ChatGPT and Claude, builds a matrix weighing value against development time, and moves forward with whatever comes out of that conversation. &#8220;It&#8217;s what I have and can keep up with, so that&#8217;s what I go with,&#8221; she said.</p><p>It&#8217;s not an isolated case. A larger batch of interviews showed the same pattern recurring often enough to become its own research question: what are the limits of using a generic AI to simulate what a customer would think?</p><p>Our suspicion, which we still need to validate, is that an AI with no context on your product and your customer base gives back what sounds plausible, not what the customer would actually say. It looks like research. It isn&#8217;t.</p><h2>Where that leaves us</h2><p>The conclusion ReveLumi itself surfaced in its own full report is that ReveLumi doesn&#8217;t compete with other research tools. It competes with the absence of research and with improvisation.</p><p>That means the right pitch isn&#8217;t &#8220;you need one more research tool&#8221;. It&#8217;s &#8220;you know talking to customers matters, but you can&#8217;t do it in a structured way&#8221;.</p><p>If that&#8217;s your day to day, it&#8217;s worth seeing what changes when the conversation with the customer happens continuously, in a structured way, without someone having to stop everything to run a formal study.</p><p><a href="https://calendar.google.com/calendar/appointments/schedules/AcZssZ1l4U38tikGBmiJWc8jkB2wkj20Ev5VlH94cY88GZOGWZ3ZKG4R9gSjj3EFt__ovsBGkUZnggcH">Book a demo to see ReveLumi</a></p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[ReveLumi compete com a ausência de pesquisa]]></title><description><![CDATA[Estamos competindo com a falta de h&#225;bito de falar com clientes &#129335;&#8205;&#9794;&#65039;]]></description><link>https://blog.revelumi.com/p/revelumi-compete-com-a-ausencia-de</link><guid isPermaLink="false">https://blog.revelumi.com/p/revelumi-compete-com-a-ausencia-de</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 03 Aug 2026 18:49:30 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Realizamos uma pesquisa simples sobre como as pessoas tomam decis&#245;es em temas relacionados ao cliente. Tivemos mais de 40 respostas.</p><p>A maioria decide primeiro com base no roadmap, nos OKRs e no que a lideran&#231;a j&#225; definiu. Quando falta informa&#231;&#227;o sobre o cliente, a resposta quase nunca &#233; uma pesquisa. &#201; marcar uma reuni&#227;o de 20 minutos, sentar do lado de um colega ou perguntar para quem est&#225; mais perto do atendimento. Pesquisa qualitativa estruturada praticamente n&#227;o existe na rotina.</p><p>Isso muda a pergunta que a gente vinha fazendo. N&#227;o &#233; &#8220;como convencer algu&#233;m a trocar a ferramenta de pesquisa que j&#225; usa?&#8221;. &#201; &#8220;como competir com o improviso e a falta de h&#225;bito de falar com a cliente?&#8221;.</p><h2>O Pix que n&#227;o estava quebrado</h2><p>Um dos relatos mais ricos veio de uma product owner que cuida de Pix em uma institui&#231;&#227;o de pagamento. O time levantou as dez maiores insatisfa&#231;&#245;es relatadas por clientes, e uma delas era recorrente: &#8220;eu n&#227;o consigo fazer Pix, nem cadastrar chave&#8221;.</p><p>O primeiro instinto de qualquer executivo, ela contou, &#233; achar que o app est&#225; com bug. S&#243; que ao cruzar os tickets com uma ferramenta que centraliza os canais de atendimento, ela viu outra coisa: cerca de 80% desses clientes tinham alguma fraude registrada em nome deles, em qualquer banco, n&#227;o s&#243; no pr&#243;prio. Uma regra do Bacen de fevereiro deste ano bloqueia o uso de Pix nesses casos, e o bloqueio &#233; o mesmo em qualquer institui&#231;&#227;o.</p><p>O dado resolveu o mist&#233;rio. Mas a pr&#243;pria PO admite que o dado n&#227;o &#233; suficiente. &#8220;No geral, &#233; muito dif&#237;cil entender o que o cliente pensa&#8221;, ela disse. Depois de explicar o problema t&#233;cnico, ainda falta a parte qualitativa: como o cliente reage ao descobrir isso, o que ele espera que o banco fa&#231;a, se a comunica&#231;&#227;o atual est&#225; clara. O dado quantitativo respondeu o qu&#234;. Ningu&#233;m tinha respondido o porqu&#234;.</p><h2>&#8220;Ele acha que somos pastelaria&#8221;</h2><p>Nem toda resposta veio em forma de caso completo. Uma PM que acabou de entrar na empresa contou que pergunta para os colegas mais antigos quando falta contexto sobre o cliente. Perguntamos se sobrava espa&#231;o para entender o cliente antes de decidir, com a press&#227;o de entrega que ela descreveu.</p><p>A resposta foi uma frase s&#243;: &#8220;n&#227;o sobra, pois acham que somos pastelaria&#8221;.</p><p>N&#227;o precisa de mais explica&#231;&#227;o. Pedido, entregue, pr&#243;ximo. &#201; a cultura que a maioria das empresas de produto vive, mesmo aquelas que dizem que colocam o cliente no centro.</p><h2>O ChatGPT n&#227;o substitui o cliente</h2><p>Uma fundadora solo relatou outro atalho comum. Quando n&#227;o tem um cliente neutro para consultar, ela conversa com ChatGPT e Claude, monta uma matriz de valor contra tempo de desenvolvimento e segue com o que sobrou dessa conversa. &#8220;&#201; o que eu tenho e consigo manter, e com isso que vou&#8221;, ela disse.</p><p>N&#227;o &#233; um caso isolado. Uma leva maior de entrevistas mostrou o mesmo padr&#227;o se repetindo com for&#231;a suficiente para virar uma pergunta de pesquisa pr&#243;pria: quais os limites de usar uma IA gen&#233;rica para simular o que uma cliente pensaria?</p><p>A suspeita que temos, e que ainda vamos validar, &#233; que uma IA sem contexto do seu produto e da sua base de clientes devolve o que soa plaus&#237;vel, n&#227;o o que a cliente de fato diria. Parece pesquisa, mas n&#227;o &#233;.</p><h2>Onde isso nos deixa</h2><p>A conclus&#227;o que a pr&#243;pria ReveLumi trouxe no seu relat&#243;rio completo &#233; que a ReveLumi n&#227;o compete com outras ferramentas de pesquisa. Compete com a aus&#234;ncia de pesquisa e com o improviso.</p><p>Isso significa que o argumento certo n&#227;o &#233; &#8220;voc&#234; precisa de mais uma ferramenta de pesquisa&#8221;. &#201; &#8220;voc&#234; sabe que falar com clientes &#233; importante, mas n&#227;o consegue fazer isso de forma estruturada&#8221;.</p><p>Se esse &#233; o seu dia a dia, vale ver o que muda quando a conversa com a cliente acontece de forma cont&#237;nua, estruturada e sem depender de algu&#233;m parar tudo para rodar uma pesquisa formal.</p><p><a href="https://calendar.google.com/calendar/appointments/schedules/AcZssZ1l4U38tikGBmiJWc8jkB2wkj20Ev5VlH94cY88GZOGWZ3ZKG4R9gSjj3EFt__ovsBGkUZnggcH">Marque uma demo para conhecer a ReveLumi</a></p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[ReveLumi isn't SaaS, it's SaS]]></title><description><![CDATA[A lesson about category that changed how we're going to talk about ReveLumi from now on.]]></description><link>https://blog.revelumi.com/p/revelumi-isnt-saas-its-sas</link><guid isPermaLink="false">https://blog.revelumi.com/p/revelumi-isnt-saas-its-sas</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 27 Jul 2026 11:34:04 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!Q5Vv!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/revelumi-nao-e-saas-e-sas">Portuguese version</a>]</p><p>Over the past few weeks, we&#8217;ve been presenting ReveLumi as a qualitative research tool. A good tool, fast, with AI. And that&#8217;s how, without meaning to, we put ReveLumi in the wrong category, and made whoever was listening misunderstand what we were actually building.</p><h3>The wrong category</h3><p>When someone hears &#8220;research tool,&#8221; their mind automatically compares it to other research tools. Dovetail, Qualtrics, User Interviews. And a new tool competes with a tool that already exists, or with however the person already solves this today, so it becomes easy to say &#8220;interesting, I&#8217;ll look into it later.&#8221;</p><p>We saw this reaction from people who already work with research inside companies. Real interest, some agreed to test it, and disappeared afterward. Our hypothesis is that what was missing was what actually competes with ReveLumi: a decision stuck because of a lack of customer information. As long as we presented ReveLumi as a tool, we were competing with other tools.</p><p>Except ReveLumi isn&#8217;t a tool that helps you do research work. It does the research work.</p><h3>SaaS helps you do it. SaS does it.</h3><p>A new category of products is emerging, called Service as Software, and it's gaining traction in the market. According to <a href="https://www.ycombinator.com/rfs#ai-native-service-companies">Y Combinator</a>:</p><blockquote><p>&#8220;Historically, services became SaaS software. More recently, they became AI copilots. [...] What excites us now is the next step: AI-native companies that don&#8217;t sell software, they sell the service. Instead of giving you a tool, they simply do the work.&#8221;</p></blockquote><p>SaaS (Software as a Service) gives you a tool to do the task faster. In the case of qualitative research, you still need to recruit, schedule, interview, and analyze, just with software helping at each step.</p><p>SaS (Service as Software), on the other hand, doesn&#8217;t give you a tool to help you do it. It does the work for you. You don&#8217;t need to know how to conduct a qualitative interview. ReveLumi does.</p><div class="captioned-image-container"><figure><a class="image-link image2" target="_blank" href="https://substackcdn.com/image/fetch/$s_!Q5Vv!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 424w, https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 848w, https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 1272w, https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png" width="1292" height="308" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:308,&quot;width&quot;:1292,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:92577,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/208672639?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 424w, https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 848w, https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 1272w, https://substackcdn.com/image/fetch/$s_!Q5Vv!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4ba759ae-f722-4bb9-aa24-8cda199ea76d_1292x308.png 1456w" sizes="100vw" loading="lazy"></picture><div></div></div></a></figure></div><p>You can already see this elsewhere. Sierra doesn&#8217;t sell support software for a customer service team to operate; its agent talks directly to the end customer and resolves the issue, and the company charges per resolved ticket, not per license.</p><p>Another important point is the pricing model. Instead of charging per user using the tool, SaS companies charge per completed task.</p><p>ReveLumi, without us ever naming it, always charged the same way. We don&#8217;t charge per user accessing a dashboard. We charge per interview with insight, the completed task. Our pricing model already showed we were SaS.</p><p>Curiously, this isn&#8217;t the first time we&#8217;ve done this without naming it. An earlier product of ours, a virtual astrologer, was also SaS: the person didn&#8217;t use a tool to learn how to interpret their own birth chart. They talked with an agent that provided the birth chart interpretation service and answered their questions.</p><p>An interesting thing to watch in the market is that some SaS companies still charge per-user subscriptions, like Lovable and Replit. You describe what you want, and the agent writes the code on its own, no programming required from you. But the pricing is still half SaaS: a fixed monthly subscription, plus a layer of usage-based credits. And that credit layer is even murkier, because it doesn't charge per completed task, it charges per computing power consumed. It's as if Netflix charged you a variable rate for the electricity you use while watching shows and movies.</p><h3>What changes when you&#8217;re offering a SaS</h3><p>This changes the question a person asks to decide whether they need you, well beyond semantics.</p><p>As SaaS, the question is &#8220;will this tool help me do my work better and faster?&#8221;</p><p>As SaS, the question becomes &#8220;is it worth doing this myself, hiring someone to do it, or letting the agent do it?&#8221;</p><p>In ReveLumi&#8217;s case, we should think about the alternative to not using our agent: the person&#8217;s own time, or the cost of hiring someone.</p><p>We noticed that this shift from SaaS to SaS also changes who feels the pain. Whoever already has a structured research process, a ResearchOps team, for example, tends to see us as a tool, and a new tool competes with one that already works. Whoever struggles to do research with users because of all the difficulties of interviewing them probably resents having to do that work, not lacking a better tool to do work they already do. For that person, the question was maybe always: who can do this for me?</p><h3>The lesson</h3><p>We thought we were competing with research tools. We were actually competing with not doing research at all, or with hiring someone to do it.</p><p>That&#8217;s what we&#8217;re going to start saying more clearly from now on: ReveLumi does the research. It&#8217;s not software that helps you do it.</p><div><hr></div><p><em><a href="https://revelumi.com/">ReveLumi</a> conducts qualitative interviews over WhatsApp, start to finish, without you needing to recruit, schedule, or analyze anything yourself.</em></p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption"> Thanks for reading Listen to Decide!   Subscribe for free to receive new posts.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[ReveLumi não é SaaS, é SaS]]></title><description><![CDATA[Um aprendizado sobre categoria que mudou como vamos falar da ReveLumi daqui pra frente.]]></description><link>https://blog.revelumi.com/p/revelumi-nao-e-saas-e-sas</link><guid isPermaLink="false">https://blog.revelumi.com/p/revelumi-nao-e-saas-e-sas</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 27 Jul 2026 11:30:06 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!YDaB!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/revelumi-isnt-saas-its-sas">Vers&#227;o em ingl&#234;s</a>]</p><p>Nas &#250;ltimas semanas, apresentamos a ReveLumi como uma ferramenta de pesquisa qualitativa. Uma boa ferramenta, r&#225;pida, com IA. E foi assim, sem querer, que colocamos a ReveLumi na categoria errada, e fizemos quem nos ouvia entender de forma equivocada o que est&#225;vamos construindo.</p><h3>A categoria errada</h3><p>Quando algu&#233;m ouve &#8220;ferramenta de pesquisa&#8221;, a cabe&#231;a compara automaticamente com outras ferramentas de pesquisa. Dovetail, Qualtrics, User Interviews. E ferramenta nova compete com ferramenta que j&#225; existe ou com o jeito que a pessoa j&#225; resolve isso hoje, ent&#227;o vira mais f&#225;cil dizer &#8220;interessante, vou ver depois&#8221;.</p><p>Vimos essa rea&#231;&#227;o com quem j&#225; trabalha com pesquisa dentro das empresas. Interesse real, algumas toparam testar, e sumiram depois. Nossa hip&#243;tese &#233; que faltou o que compete de verdade com a ReveLumi: a decis&#227;o travada por falta de informa&#231;&#227;o de cliente. Enquanto apresentamos a ReveLumi como ferramenta, competimos com outras ferramentas. </p><p>S&#243; que a ReveLumi n&#227;o &#233; uma ferramenta que ajuda a fazer o trabalho de pesquisa. Ela faz o trabalho de pesquisa.</p><h3>SaaS ajuda a fazer. SaS faz.</h3><p>Est&#225; surgindo uma nova categoria de produtos, batizada de Service as Software, que vem ganhando for&#231;a no mercado. Segundo a <a href="https://www.ycombinator.com/rfs#ai-native-service-companies">Y Combinator</a>:</p><blockquote><p>&#8220;Historicamente, servi&#231;os viraram software SaaS. Mais recentemente, viraram copilotos de IA. [...] O que nos empolga agora &#233; o pr&#243;ximo passo: empresas nativas de IA que n&#227;o vendem software, vendem o servi&#231;o. Em vez de te dar uma ferramenta, elas simplesmente fazem o trabalho.&#8221;</p></blockquote><p>SaaS (Software as a Service) te d&#225; uma ferramenta para voc&#234; fazer a tarefa mais r&#225;pido. No caso da pesquisa qualitativa, voc&#234; precisa recrutar, agendar, entrevistar, e analisar, s&#243; que com um software ajudando em cada etapa.</p><p>Por outro lado, com um SaS (Service as Software), em vez de uma ferramenta para ajudar a fazer, ele faz o trabalho por voc&#234;. Voc&#234; n&#227;o precisa saber conduzir uma entrevista qualitativa. A ReveLumi conduz.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!YDaB!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!YDaB!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 424w, https://substackcdn.com/image/fetch/$s_!YDaB!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 848w, https://substackcdn.com/image/fetch/$s_!YDaB!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 1272w, https://substackcdn.com/image/fetch/$s_!YDaB!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!YDaB!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png" width="1292" height="342" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:342,&quot;width&quot;:1292,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:104445,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/208667141?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!YDaB!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 424w, https://substackcdn.com/image/fetch/$s_!YDaB!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 848w, https://substackcdn.com/image/fetch/$s_!YDaB!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 1272w, https://substackcdn.com/image/fetch/$s_!YDaB!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F545ffdca-fcb0-4cfc-a423-a551b20ecb14_1292x342.png 1456w" sizes="100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>J&#225; d&#225; para ver isso em outros lugares. A Sierra n&#227;o vende um software de atendimento para o time de suporte operar; a agente dela conversa diretamente com o cliente final e resolve o problema, e a empresa cobra por ticket resolvido, n&#227;o por licen&#231;a de acesso.</p><p>Outro ponto importante &#233; a forma de cobran&#231;a. Em vez de cobrar por usu&#225;ria utilizando a ferramenta, os SaS cobram por tarefa conclu&#237;da.</p><p>A ReveLumi, sem a gente ter nomeado isso, sempre cobrou do mesmo jeito. N&#227;o cobramos por usu&#225;ria que acessa o painel. Cobramos por entrevista com insight, que &#233; a tarefa conclu&#237;da. Nosso modelo de cobran&#231;a j&#225; mostrava que somos SaS.</p><p>Curiosamente, essa n&#227;o &#233; a primeira vez que a gente faz isso sem nomear. Um produto anterior nosso, uma astr&#243;loga virtual, tamb&#233;m era um SaS: a pessoa n&#227;o usava uma ferramenta para aprender a interpretar o pr&#243;prio mapa astral. Ela conversava com uma agente que prestava o servi&#231;o de interpreta&#231;&#227;o do mapa astral e respondia &#224;s perguntas. </p><p>Um ponto interessante para observarmos no mercado &#233; que h&#225; SaS que s&#227;o cobrados por assinatura por usu&#225;rio, como Lovable e Replit. Voc&#234; descreve o que quer, e o agente escreve o c&#243;digo sozinho, sem voc&#234; precisar programar. Mas a cobran&#231;a ainda &#233; meio SaaS: assinatura mensal fixa, mais uma camada de cr&#233;ditos por uso. E essa camada de cr&#233;ditos &#233; ainda mais nebulosa, pois n&#227;o cobra por tarefa feita, mas sim por poder computacional consumido. &#201; como se a Netflix te cobrasse vari&#225;vel pela quantidade de energia el&#233;trica consumida enquanto voc&#234; assiste a s&#233;ries e filmes.</p><h3>O que muda quando voc&#234; est&#225; oferecendo um SaS</h3><p>Isso muda a pergunta que a pessoa faz para decidir se precisa de voc&#234;, bem al&#233;m da sem&#226;ntica.</p><p>Como SaaS, a pergunta &#233; &#8220;essa ferramenta vai me ajudar a fazer o meu trabalho melhor e mais r&#225;pido?&#8221;. </p><p>Como SaS, a pergunta vira &#8220;vale mais a pena eu mesma fazer isso, contratar algu&#233;m para fazer, ou deixar a agente fazer?&#8221;. </p><p>No caso da ReveLumi, devemos pensar sobre a alternativa de n&#227;o usar nossa agente: o pr&#243;prio tempo da pessoa, ou o custo de contratar.</p><p>Percebemos que essa mudan&#231;a de SaaS para SaS tamb&#233;m muda quem sente a dor. Quem j&#225; tem um processo de pesquisa estruturado, um time de ResearchOps, por exemplo, tende a nos ver como ferramenta, e uma ferramenta nova compete com uma que j&#225; funciona. J&#225; quem sofre para fazer pesquisas com usu&#225;rias pelas dificuldades todas do trabalho de entrevist&#225;-las, provavelmente se ressente de ter que fazer esse trabalho, n&#227;o de uma ferramenta melhor para fazer um trabalho que j&#225; faz. Para essa pessoa, a pergunta talvez sempre tenha sido: quem pode fazer isso por mim?</p><h3>O aprendizado</h3><p>Ach&#225;vamos que est&#225;vamos competindo com ferramentas de pesquisa. Est&#225;vamos, na real, competindo com n&#227;o fazer pesquisa nenhuma ou com contratar algu&#233;m para fazer.</p><p>&#201; isso que vamos passar a falar mais claramente daqui pra frente: a ReveLumi faz a pesquisa. N&#227;o &#233; um software que ajuda voc&#234; a fazer.</p><div><hr></div><p><em><a href="https://revelumi.com/">ReveLumi</a> conduz entrevistas qualitativas por WhatsApp, do in&#237;cio ao fim, sem voc&#234; precisar recrutar, agendar ou analisar nada sozinha. </em></p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[The NPS score tells only a small part of the story]]></title><description><![CDATA[We tested ReveLumi's conversational NPS by asking 133 people what they think of LinkedIn.]]></description><link>https://blog.revelumi.com/p/the-nps-score-tells-only-a-small</link><guid isPermaLink="false">https://blog.revelumi.com/p/the-nps-score-tells-only-a-small</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 20 Jul 2026 15:26:42 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!5N3Q!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/a-nota-do-nps-conta-so-um-pedaco">Portuguese version</a>]</p><p>NPS is the easiest metric to run and the hardest to actually use. Everyone sends the same question, 0 to 10, and gets a number back. The number tells you if things are going well or badly. It doesn&#8217;t tell you why.</p><p>The open text field in traditional NPS was supposed to solve that, but in practice it becomes a graveyard of one-line answers. &#8220;Good.&#8221; &#8220;Could be better.&#8221; &#8220;Not sure.&#8221; Nobody elaborates because nobody asks them to.</p><p>That&#8217;s why we built conversational NPS into ReveLumi. The person gives a score, and the agent talks with them to understand the why, actually digging in, not stopping at the first answer. To test it, we ran a survey about a product everyone has an opinion on: LinkedIn.</p><h3>The number</h3><p>133 responses. NPS of +7. 34% promoters, 39% passives, 27% detractors. Here&#8217;s the report building itself in real time as responses come in:</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!5N3Q!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!5N3Q!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 424w, https://substackcdn.com/image/fetch/$s_!5N3Q!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 848w, https://substackcdn.com/image/fetch/$s_!5N3Q!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 1272w, https://substackcdn.com/image/fetch/$s_!5N3Q!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!5N3Q!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png" width="1456" height="1278" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1278,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:286426,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/207790827?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!5N3Q!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 424w, https://substackcdn.com/image/fetch/$s_!5N3Q!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 848w, https://substackcdn.com/image/fetch/$s_!5N3Q!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 1272w, https://substackcdn.com/image/fetch/$s_!5N3Q!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6f192462-59b8-4aa0-a0d4-3a1d4c154ddc_1568x1376.png 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>An NPS of +7, on its own, isn&#8217;t much. It says LinkedIn is more &#8220;tolerated&#8221; than &#8220;loved.&#8221; But that&#8217;s all it says. It doesn&#8217;t say what sustains the score, what brings it down, or what would change someone&#8217;s mind. That&#8217;s where the number stops and the conversation starts.</p><h3>What sustains the score, and what brings it down</h3><p>The biggest positive factor, by far, was professional networking: expanding your network, gaining visibility with recruiters, being found.</p><p>The biggest negative factor wasn&#8217;t price, or a bug, or the interface. It was content. Specifically: the feed being flooded with bots and AI-generated content, and right behind it, a term one respondent brought into the conversation herself:</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!3b8U!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!3b8U!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 424w, https://substackcdn.com/image/fetch/$s_!3b8U!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 848w, https://substackcdn.com/image/fetch/$s_!3b8U!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 1272w, https://substackcdn.com/image/fetch/$s_!3b8U!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!3b8U!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png" width="1456" height="930" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:930,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:497106,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/207790827?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!3b8U!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 424w, https://substackcdn.com/image/fetch/$s_!3b8U!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 848w, https://substackcdn.com/image/fetch/$s_!3b8U!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 1272w, https://substackcdn.com/image/fetch/$s_!3b8U!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2b5af6ed-9855-4632-943b-68de4eda9aad_2092x1336.png 1456w" sizes="100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><blockquote><p>&#8220;There&#8217;s even a term for it, LinkedDisney. It&#8217;s because there are these very forced testimonials.&#8221;</p></blockquote><p>LinkedDisney. A nickname already circulating among managers and directors to describe the feed&#8217;s performative exaggeration: overly dramatic stories, overly grand achievements, everything with a forced happy ending. No closed-ended question would ever land on that term, and no automated sentiment report would invent it. It only surfaces in a conversation, because someone needs to say, in their own words, what they actually feel. And the report doesn&#8217;t just capture the phrase, it weighs how much it contributes to the score, how many people touched on it, and links back to every original conversation.</p><h3>The irony we didn&#8217;t plan for</h3><p>We ran this survey with an AI agent. And the biggest villain the survey found was soulless AI-generated content.</p><p>It&#8217;s not a contradiction, it&#8217;s the distinction that matters most right now. There&#8217;s AI that produces empty volume, text that only exists to fill space and look productive. And there&#8217;s AI that listens, asks the follow-up question, and gives back real understanding. The difference isn&#8217;t in the technology. It&#8217;s in what it&#8217;s used for: generating more content, or generating more understanding.</p><p>Conversational NPS is a bet on the second one.</p><h3>What the feature delivers</h3><p>Instead of an open field most people ignore, the agent asks back, adapts the follow-up to the score (detractor, passive, promoter), and doesn&#8217;t let the person off with a shallow first answer. The result, as you saw above, isn&#8217;t just a number. It&#8217;s a map of factors, weighted and counted, linked to real conversations you can reopen anytime.</p><p>A traditional NPS would tell you: &#8220;+7, driven by networking.&#8221; Ours tells you that, and also hands you &#8220;LinkedDisney,&#8221; named by a real person, with enough weight to explain why the score isn&#8217;t higher.</p><p>Data tells us what happened. Customers tell us why. Now, with conversational NPS, the NPS tells us too.</p><div><hr></div><p><em>See it for yourself: try <a href="https://revelumi.com/c/H08GABDR?lang=en">LinkedIn&#8217;s NPS survey</a>.</em></p><p><em>Want to test conversational NPS on your own product? Book a <a href="https://calendar.app.google/u7eprGHcvebAk8AaA">demo session</a> with the founders.</em></p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[A nota do NPS conta só um pedaço da história]]></title><description><![CDATA[Testamos o NPS conversacional da ReveLumi perguntando a 133 pessoas o que elas acham do LinkedIn.]]></description><link>https://blog.revelumi.com/p/a-nota-do-nps-conta-so-um-pedaco</link><guid isPermaLink="false">https://blog.revelumi.com/p/a-nota-do-nps-conta-so-um-pedaco</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 20 Jul 2026 13:37:25 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!7VwS!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/the-nps-score-tells-only-a-small">Vers&#227;o em ingl&#234;s</a>]</p><p>NPS &#233; a m&#233;trica mais f&#225;cil de rodar e a mais dif&#237;cil de aproveitar. Todo mundo manda a mesma pergunta, de 0 a 10, e recebe de volta um n&#250;mero. O n&#250;mero diz se est&#225; indo bem ou mal. N&#227;o diz por qu&#234;.</p><p>O campo aberto do NPS tradicional devia resolver isso, mas na pr&#225;tica vira um cemit&#233;rio de respostas de uma linha. &#8220;Bom.&#8221; &#8220;Poderia melhorar.&#8221; &#8220;N&#227;o sei.&#8221; Ningu&#233;m elabora porque ningu&#233;m pergunta de volta.</p><p>Por isso constru&#237;mos o NPS conversacional na ReveLumi. A pessoa d&#225; a nota, e a agente conversa para entender o porqu&#234;, aprofundando de verdade, n&#227;o parando na primeira resposta. Para testar, rodamos uma pesquisa sobre um produto que todo mundo tem opini&#227;o: o LinkedIn.</p><h3>O n&#250;mero</h3><p>133 respostas. NPS de +7. 34% promotoras, 39% neutras, 27% detratoras. Assim que o relat&#243;rio se monta sozinho, em tempo real, conforme as respostas chegam:</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!7VwS!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!7VwS!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 424w, https://substackcdn.com/image/fetch/$s_!7VwS!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 848w, https://substackcdn.com/image/fetch/$s_!7VwS!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 1272w, https://substackcdn.com/image/fetch/$s_!7VwS!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!7VwS!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png" width="1456" height="1318" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1318,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:283076,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/207775274?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!7VwS!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 424w, https://substackcdn.com/image/fetch/$s_!7VwS!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 848w, https://substackcdn.com/image/fetch/$s_!7VwS!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 1272w, https://substackcdn.com/image/fetch/$s_!7VwS!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6ec75c11-6e7b-4963-87b9-e090afdd3223_1524x1380.png 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>Um NPS de +7, por si s&#243;, &#233; pouco. Diz que o LinkedIn est&#225; mais para &#8220;tolerado&#8221; do que &#8220;amado&#8221;. Mas &#233; s&#243; isso que diz. N&#227;o diz o que sustenta a nota, nem o que a derruba, nem o que faria a pessoa mudar de opini&#227;o. &#201; aqui que o n&#250;mero para e a conversa come&#231;a.</p><h3>O que sustenta, e o que derruba, a nota</h3><p>O maior fator positivo, de longe, foi a oportunidade de conex&#227;o profissional: expandir a rede, ganhar visibilidade para recrutadores, ser encontrado.</p><p>O maior fator negativo n&#227;o foi pre&#231;o, nem bug, nem interface. Foi conte&#250;do. Mais especificamente: a polui&#231;&#227;o do feed por bots e conte&#250;do gerado por IA, e logo atr&#225;s, um termo que uma respondente trouxe para a conversa:</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!pIRc!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!pIRc!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 424w, https://substackcdn.com/image/fetch/$s_!pIRc!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 848w, https://substackcdn.com/image/fetch/$s_!pIRc!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 1272w, https://substackcdn.com/image/fetch/$s_!pIRc!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!pIRc!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png" width="1456" height="913" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/bb49512d-499c-4621-abfd-a24658d06567_2096x1314.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:913,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:463997,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://listentodecide.substack.com/i/207775274?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!pIRc!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 424w, https://substackcdn.com/image/fetch/$s_!pIRc!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 848w, https://substackcdn.com/image/fetch/$s_!pIRc!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 1272w, https://substackcdn.com/image/fetch/$s_!pIRc!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb49512d-499c-4621-abfd-a24658d06567_2096x1314.png 1456w" sizes="100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><blockquote><p>&#8220;Tem at&#233; um termo que se chama LinkedDisney. &#201; porque tem l&#225; depoimentos muito for&#231;ados.&#8221;</p></blockquote><p>LinkedDisney. Um apelido que j&#225; circula em rodas de ger&#234;ncia e diretoria para descrever o exagero perform&#225;tico do feed: hist&#243;rias dram&#225;ticas demais, conquistas grandiosas demais, tudo com brilho de final feliz. Nenhuma pergunta fechada chegaria nesse termo, nem um relat&#243;rio autom&#225;tico de sentimento o inventaria. Ele s&#243; aparece numa conversa, porque algu&#233;m precisa contar, com as pr&#243;prias palavras, o que sente. E o relat&#243;rio n&#227;o s&#243; registra a frase, ele pesa quanto ela contribui para a nota, quantas pessoas tocaram nesse ponto, e liga de volta para cada conversa original.</p><h3>A ironia que n&#227;o planejamos</h3><p>Rodamos essa pesquisa com uma agente de IA. E o maior vil&#227;o que a pesquisa encontrou foi conte&#250;do gerado por IA sem alma.</p><p>N&#227;o &#233; contradi&#231;&#227;o, &#233; a distin&#231;&#227;o que mais importa agora. Existe IA que produz volume vazio, textos que s&#243; existem para preencher espa&#231;o e parecer produtivo. E existe IA que escuta, faz a pergunta seguinte, e devolve entendimento real. A diferen&#231;a n&#227;o est&#225; na tecnologia. Est&#225; em para que ela &#233; usada: gerar mais conte&#250;do, ou gerar mais compreens&#227;o.</p><p>O NPS conversacional &#233; uma aposta na segunda.</p><h3>O que a funcionalidade entrega</h3><p>Em vez de um campo aberto que a maioria ignora, a agente pergunta de volta, adapta a pergunta conforme a nota (detratora, neutra, promotora), e n&#227;o solta a pessoa na primeira resposta rasa. O resultado, como voc&#234; viu acima, n&#227;o &#233; s&#243; um n&#250;mero. &#201; um mapa de fatores, com peso e volume, ligado a conversas reais que voc&#234; pode reabrir a qualquer momento.</p><p>Um NPS tradicional te diria: &#8220;+7, com destaque para networking.&#8221; O nosso te diz isso, e tamb&#233;m te entrega &#8220;LinkedDisney&#8221;, nomeado por gente de carne e osso, com peso suficiente para explicar por que a nota n&#227;o &#233; maior.</p><p>Dados dizem o qu&#234;. Clientes dizem o porqu&#234;. Agora, com o NPS conversacional, o NPS tamb&#233;m diz.</p><div><hr></div><p><em>Experimente voc&#234; mesmo a <a href="https://revelumi.com/c/H08GABDR">pesquisa de NPS do LinkedIn</a>.</em></p><p><em>Quer testar o NPS conversacional no seu produto? Marque <a href="https://calendar.app.google/u7eprGHcvebAk8AaA">uma sess&#227;o de demo</a> com os founders.</em></p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[What people tell you when someone finally asks]]></title><description><![CDATA[159 conversations about Netflix, and what they reveal about the value of listening to customers.]]></description><link>https://blog.revelumi.com/p/what-people-tell-you-when-someone</link><guid isPermaLink="false">https://blog.revelumi.com/p/what-people-tell-you-when-someone</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 13 Jul 2026 21:15:41 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/o-que-as-pessoas-contam-quando-alguem">Portuguese version</a>]</p><p>Last week I wrote about why we spend twenty minutes scrolling Netflix and leave without watching anything. The post came out of 159 conversations I ran over WhatsApp, with ReveLumi.</p><p>There was material I didn&#8217;t use, and it&#8217;s more interesting than the diagnosis. Alongside the complaints, people offered suggestions. Nobody asked them to. They simply described what they wanted, with the clarity of someone who lives with the problem every day.</p><p>No ideation workshop, no sticky notes. Just open questions and people talking.</p><h3>What they told us</h3><p><strong>Ask us.</strong> This was the most recurring request, and the most disconcerting.</p><blockquote><p>&#8220;There could be a way to talk to the platform so it understands what you&#8217;d like to watch that day.&#8221;</p><p>&#8220;I think it could ask me questions, or ask me to rate what I&#8217;ve already watched. Today it&#8217;s just thumbs up or thumbs down.&#8221;</p><p>&#8220;Netflix&#8217;s algorithm doesn&#8217;t offer me things I like. There could be a minimal questionnaire so the offering is more accurate.&#8221;</p></blockquote><p>People are asking, in plain words, to be asked. They notice that the product tries to guess what they want, gets it wrong, and never does the simplest thing: ask.</p><p><strong>Bring the critics inside.</strong> Since they leave the platform to check a rating before deciding, the suggestion came ready-made:</p><blockquote><p>&#8220;A great idea would be for Netflix to bring in a list of films and series ranked by the main critic sites, like IMDb.&#8221;</p></blockquote><p><strong>Give us control.</strong></p><blockquote><p>&#8220;Let the user build their own filters.&#8221;</p><p>&#8220;I&#8217;d change how the offerings are laid out, I&#8217;d create other categories, since the catalog is enormous.&#8221;</p></blockquote><p><strong>Consider my moment.</strong></p><blockquote><p>&#8220;Since it depends on the mood of the day, maybe something related to that?&#8221;</p></blockquote><h3>The answer that floored me</h3><p>In the middle of the conversations, someone answered the question about what would solve their problem on Netflix:</p><blockquote><p>&#8220;Yes, a chat or conversational AI agent like you would help.&#8221;</p></blockquote><p>Read it again, with the context: she was being interviewed by a conversational AI agent. And she told the agent that an agent would solve her problem. She lived the solution in the very moment she described it.</p><p>She felt, in the conversation itself, the difference between being guessed at and being asked. And she knew how to name it.</p><h3>No dashboard would produce this list</h3><p>Think about what Netflix has. The full click history, screen time, completion rates, A/B tests running all the time. And think about what the data would say about these requests.</p><p>The data would show that people spend a long time on the browsing screen. It wouldn&#8217;t say that they want to be asked. The data would show that they abandon the session. It wouldn&#8217;t say that they went to check the IMDb rating and never came back. The data would show a drop in engagement on a Tuesday night. It wouldn&#8217;t say that the person was exhausted and wanted something light.</p><p>All of this came out of conversations that lasted a few minutes, over two days, with no interviews scheduled.</p><h3>This is not a roadmap</h3><p>And here comes the part that matters, especially for people who build products.</p><p>None of what&#8217;s above is a list of things to build. The conversation doesn&#8217;t give you the answer. It brings us closer to the problem, and sometimes arrives with suggestions for a solution. But the responsibility to extract the insight, and then to find the solution that makes sense, still belongs to the person who builds the product.</p><p>When someone asks for &#8220;a minimal questionnaire,&#8221; the value isn&#8217;t in the questionnaire. It&#8217;s in what it reveals: this person wants to be heard instead of guessed at. The right solution could be something else entirely, and finding out which one takes solution discovery, testing, measuring, learning.</p><p>Remember what we saw in the post about ROI: about two thirds of product ideas fail to improve the metric they were meant to improve. Talking to customers doesn&#8217;t change that statistic. What changes is the quality of what goes into the testing queue. A hypothesis born from a real pain, heard from dozens of people, has a better chance of landing in the third that works than a hunch from a meeting room.</p><p>And in the end, the product work stays the same as always: find the solution that solves the customer&#8217;s problem and delivers results for the company. The conversation doesn&#8217;t replace that work. It gives you better raw material to do it.</p><h3>And your customers?</h3><p>They live with your product every day. They already know what&#8217;s broken, they&#8217;ve already imagined how it could be better, and they&#8217;ve probably already solved on the outside what you don&#8217;t solve on the inside.</p><p>They don&#8217;t have your roadmap. But they have the problem it&#8217;s supposed to solve. Nobody asked them.</p><p>Data tells us what happened. Customers tell us why. And it&#8217;s on us to figure out how.</p><div><hr></div><p><em>Want to see how it works? Try <a href="https://revelumi.com/c/06VP6MA2?lang=en">being interviewed by ReveLumi</a> about how you choose what to watch.</em></p><p><strong>Book a demo session with the founders</strong><br>If you&#8217;d like to learn more about ReveLumi, <a href="https://calendar.app.google/uNvL237znePbcFmBA">book a demo session</a> with Anderson and me. We&#8217;ll show you what ReveLumi does today, what we&#8217;re planning for it in the future, and answer any questions you have.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[O que as pessoas contam quando alguém finalmente pergunta]]></title><description><![CDATA[159 conversas sobre a Netflix, e o que elas revelam sobre o valor de ouvir clientes.]]></description><link>https://blog.revelumi.com/p/o-que-as-pessoas-contam-quando-alguem</link><guid isPermaLink="false">https://blog.revelumi.com/p/o-que-as-pessoas-contam-quando-alguem</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 13 Jul 2026 21:13:19 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Semana passada escrevi sobre por que a gente passa vinte minutos rolando a Netflix e sai sem assistir nada. O post nasceu de 159 conversas que rodei pelo WhatsApp, com a ReveLumi.</p><p>Ficou um material que eu n&#227;o usei, e ele &#233; mais interessante que o diagn&#243;stico. Junto com as reclama&#231;&#245;es, as pessoas ofereceram sugest&#245;es. Ningu&#233;m pediu. Elas simplesmente descreveram o que queriam, com a clareza de quem convive com o problema todo dia.</p><p>Sem workshop de idea&#231;&#227;o, sem post-it. S&#243; perguntas abertas e gente falando.</p><h3>O que elas contaram</h3><p><strong>Perguntem pra gente.</strong> Esse foi o pedido mais recorrente, e o mais desconcertante.</p><blockquote><p>&#8220;Poderia ter a possibilidade de conversar com a plataforma pra ela entender o que voc&#234; gostaria de ver naquele dia.&#8221;</p><p>&#8220;Acho que ela poderia me fazer perguntas, ou mesmo pedir pra dar notas pro que j&#225; assisti. Hoje em dia &#233; apenas curtir ou n&#227;o curtir.&#8221;</p><p>&#8220;O algoritmo da Netflix n&#227;o oferta coisas que eu gosto. Poderia ter um question&#225;rio m&#237;nimo pra que a oferta fosse mais assertiva.&#8221;</p></blockquote><p>As pessoas est&#227;o pedindo, com todas as letras, para serem perguntadas. Elas percebem que o produto tenta adivinhar o que elas querem, erra, e nunca faz a coisa mais simples: perguntar.</p><p><strong>Tragam a cr&#237;tica pra dentro.</strong> Como elas saem da plataforma para ver nota antes de decidir, a sugest&#227;o veio pronta:</p><blockquote><p>&#8220;Uma &#243;tima ideia seria a Netflix j&#225; trazer uma lista de filmes e s&#233;ries organizada pelas avalia&#231;&#245;es dos principais sites de cr&#237;tica, tipo o IMDb.&#8221;</p></blockquote><p><strong>Deixem a gente no controle.</strong></p><blockquote><p>&#8220;Permitir que o usu&#225;rio monte seus pr&#243;prios filtros.&#8221;</p><p>&#8220;Eu mudaria a disposi&#231;&#227;o das ofertas, criaria outras categorias, j&#225; que a diversifica&#231;&#227;o do servi&#231;o &#233; gigantesca.&#8221;</p></blockquote><p><strong>Considerem o meu momento.</strong></p><blockquote><p>&#8220;Como &#233; algo que depende do humor do dia, talvez algo relacionado a isso?&#8221;</p></blockquote><h3>A resposta que me deixou sem ch&#227;o</h3><p>No meio das conversas, algu&#233;m respondeu &#224; pergunta sobre o que resolveria o problema dela na Netflix:</p><blockquote><p>&#8220;Sim, um chat ou agente de IA conversacional igual voc&#234; ajudaria.&#8221;</p></blockquote><p>L&#234; de novo, com o contexto: ela estava sendo entrevistada por uma agente de IA conversacional. E disse, para a agente, que uma agente resolveria o problema dela. Ela viveu a solu&#231;&#227;o no momento em que a descreveu.</p><p>Ela sentiu, na pr&#243;pria conversa, a diferen&#231;a entre ser adivinhada e ser perguntada. E soube nomear.</p><h3>Nenhum dashboard produziria essa lista</h3><p>Pensa no que a Netflix tem. Todo o hist&#243;rico de cliques, tempo de tela, taxa de conclus&#227;o, testes A/B rodando o tempo todo. E pensa no que os dados diriam sobre esses pedidos.</p><p>O dado mostraria que as pessoas passam muito tempo na tela de navega&#231;&#227;o. Ele n&#227;o diria que elas querem ser perguntadas. O dado mostraria que elas abandonam a sess&#227;o. Ele n&#227;o diria que elas foram ver a nota no IMDb e n&#227;o voltaram. O dado mostraria queda no engajamento numa ter&#231;a &#224; noite. Ele n&#227;o diria que a pessoa estava exausta e queria algo leve.</p><p>Tudo isso saiu de conversas de poucos minutos, em dois dias, sem agendar nenhuma entrevista.</p><h3>Isso n&#227;o &#233; um roadmap</h3><p>E aqui vem a parte que importa, principalmente para quem faz produto.</p><p>Nada do que est&#225; a&#237; em cima &#233; uma lista de coisas para construir. A conversa n&#227;o d&#225; a resposta. Ela aproxima a gente do problema, e &#224;s vezes chega com sugest&#245;es de solu&#231;&#227;o. Mas a responsabilidade de extrair o insight, e depois encontrar a solu&#231;&#227;o que faz sentido, continua sendo de quem faz produto.</p><p>Quando algu&#233;m pede &#8220;um question&#225;rio m&#237;nimo&#8221;, o valor n&#227;o est&#225; no question&#225;rio. Est&#225; no que aquilo revela: essa pessoa quer ser ouvida em vez de adivinhada. A solu&#231;&#227;o certa pode ser outra coisa completamente, e descobrir qual &#233; passa por discovery de solu&#231;&#227;o, testar, medir, aprender.</p><p>Lembra do que vimos no post sobre ROI: cerca de dois ter&#231;os das ideias de produto falham em melhorar a m&#233;trica que deveriam. Conversar com cliente n&#227;o muda essa estat&#237;stica. O que muda &#233; a qualidade do que entra na fila de teste. Uma hip&#243;tese que nasceu de uma dor real, ouvida de dezenas de pessoas, tem mais chance de estar no ter&#231;o que d&#225; certo do que um palpite de sala de reuni&#227;o.</p><p>E, no fim, o trabalho de produto continua sendo o mesmo de sempre: encontrar a solu&#231;&#227;o que resolve o problema da cliente e gera resultado para a empresa. A conversa n&#227;o substitui esse trabalho. Ela d&#225; mat&#233;ria-prima melhor para faz&#234;-lo.</p><h3>E as suas clientes?</h3><p>Elas convivem com o seu produto todo dia. J&#225; sabem o que est&#225; errado, j&#225; imaginaram como seria melhor, e provavelmente j&#225; resolveram por fora aquilo que voc&#234; n&#227;o resolve por dentro.</p><p>Elas n&#227;o t&#234;m o seu roadmap. Mas t&#234;m o problema que ele deveria resolver. S&#243; ningu&#233;m perguntou.</p><p>Dados dizem o qu&#234;. Clientes dizem o porqu&#234;. E cabe &#224; gente descobrir o como.</p><div><hr></div><p><em>Quer ver como &#233;? Experimente <a href="https://revelumi.com/c/06VP6MA2">ser entrevistado pela ReveLumi</a> sobre como voc&#234; escolhe o que assistir.</em></p><p><strong>Marque uma sess&#227;o de demo com os founders</strong><br>Se voc&#234; quer conhecer mais sobre a ReveLumi, <a href="https://calendar.app.google/uNvL237znePbcFmBA">agende uma sess&#227;o de demo</a> com o Anderson e comigo para a gente te mostrar o que a ReveLumi faz hoje, o que estamos pensando para ela no futuro e para tirar todas as suas d&#250;vidas.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[Netflix knows everything about us. And we can't find anything to watch.]]></title><description><![CDATA[Netflix is one of the platforms with the most behavioral data, and we still spend twenty minutes scrolling without deciding anything.]]></description><link>https://blog.revelumi.com/p/netflix-knows-everything-about-us</link><guid isPermaLink="false">https://blog.revelumi.com/p/netflix-knows-everything-about-us</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 06 Jul 2026 09:24:11 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/a-netflix-sabe-tudo-sobre-a-gente">Portuguese version</a>]</p><p>Netflix knows every second we watched. Where we paused, where we quit, what we rewatched, at what time, on which screen, what we abandoned halfway. It&#8217;s one of the platforms with the most detailed picture of its customers&#8217; behavior that exists. And its recommendation engine is one of the most studied in the industry.</p><p>So why do we still open it, spend twenty minutes scrolling, and leave without watching anything?</p><p>I asked that question to 159 people, in conversations over WhatsApp. The answers were remarkably similar, and they all point to the same place.</p><h3>People leave Netflix to decide what to watch on Netflix</h3><p>This was the strongest pattern, and the most revealing. Plenty of people described the same ritual: open Netflix, don&#8217;t decide, go somewhere else, and sometimes come back.</p><blockquote><p>&#8220;First I go into the platform, check the suggestions and new releases, but I end up deciding with ChatGPT or some Google search result.&#8221;</p><p>&#8220;I go looking for online reviews, like IMDb. Something Netflix doesn&#8217;t offer.&#8221;</p><p>&#8220;I don&#8217;t watch a single film without knowing the critics&#8217; score first.&#8221;</p></blockquote><p>Think about what this means. One of the platforms with the most data about what we like can&#8217;t get us to hit play without us going to find the opinion of a critic, a friend, or a stranger on TikTok. The data doesn&#8217;t close the decision. A conversation with another person closes it.</p><h3>The data knows what we watched. It doesn&#8217;t know who&#8217;s on the couch.</h3><p>One answer explained, on its own, one of the biggest limitations of behavioral data:</p><blockquote><p>&#8220;I use Netflix with my boyfriend, so the algorithm gets muddled and gives me the wrong recommendations, more aimed at him than at me.&#8221;</p></blockquote><p>The history belongs to a profile. But a profile isn&#8217;t a person, it&#8217;s sometimes two, three, a whole family on a couch. The data records the clicks, it doesn&#8217;t know who&#8217;s holding the remote tonight. And no one is going to tell the algorithm &#8220;tonight it&#8217;s just me&#8221; or &#8220;tonight it&#8217;s the couple.&#8221; That would only surface if someone asked. And that&#8217;s what happened, I used ReveLumi to ask, and people answered.</p><h3>&#8220;Depends on my mood&#8221;</h3><p>That phrase, or a version of it, came up several times:</p><blockquote><p>&#8220;Depends on the day. Some days I pick anything, some days I browse more.&#8221;</p><p>&#8220;Depends a lot on my mood.&#8221;</p></blockquote><p>What we want to watch on an exhausted Tuesday after work has nothing to do with what we want on a relaxed Friday. Same person, same history, same data. What changed was the person&#8217;s state that night. And a state of mind isn&#8217;t in the click history. No amount of behavioral data captures &#8220;how you&#8217;re feeling today,&#8221; because that isn&#8217;t a past behavior, it&#8217;s a present context. Only a question reaches it.</p><h3>The cost of not knowing why</h3><p>The most interesting part is that the failure isn&#8217;t only about experience. It becomes a business problem, in the words of the people themselves:</p><blockquote><p>&#8220;It was supposed to be a moment to relax and I can&#8217;t relax. A lot of the time I end up going to other streamings.&#8221;</p><p>&#8220;I&#8217;ve given up on watching before, but the most common thing is to go find something on another streaming service.&#8221;</p></blockquote><p>When the recommendation doesn&#8217;t understand the person, they don&#8217;t keep looking forever. They leave. They go to the competitor or give up on the night. All the data in the world doesn&#8217;t prevent this, because what&#8217;s missing isn&#8217;t more data about their past. It&#8217;s understanding what they want now, and why.</p><h3>What this has to do with your product</h3><p>Netflix is the extreme case, and that&#8217;s why it teaches. If one of the companies with the most behavioral data on the planet still loses the person at the moment of deciding, what makes you think your dashboard answers every question about your product?</p><p>Data is a snapshot of what happened. It&#8217;s good at that, and we need it. But it doesn&#8217;t tell us why it happened, nor what the person felt, nor what they want now. That piece, the most important one for deciding, lives in a single place, in the conversation.</p><p>Data tells us what happened. Customers tell us why. Netflix has data to spare, and we still leave to ask a friend what to watch. Maybe it&#8217;s time to ask your customers what your data doesn&#8217;t tell you.</p><p>Want to feel it firsthand? Try <a href="https://revelumi.com/c/06VP6MA2?lang=en">being interviewed by ReveLumi</a> about how you choose what to watch on Netflix.</p><h3>Book a demo session with the founders</h3><p>If you&#8217;d like to learn more about ReveLumi, <a href="https://calendar.app.google/saPyvbntCkFALkom8">book a demo session</a> with Anderson and me. We&#8217;ll show you what ReveLumi does today, what we&#8217;re planning for it in the future, and answer any questions you have.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[A Netflix sabe tudo sobre a gente. E a gente não acha o que assistir.]]></title><description><![CDATA[Netflix &#233; uma das plataformas com mais dados de comportamento e mesmo assim a gente ainda passa vinte minutos rolando a tela sem decidir nada.]]></description><link>https://blog.revelumi.com/p/a-netflix-sabe-tudo-sobre-a-gente</link><guid isPermaLink="false">https://blog.revelumi.com/p/a-netflix-sabe-tudo-sobre-a-gente</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Sat, 04 Jul 2026 19:08:21 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>A Netflix sabe cada segundo que a gente assistiu. Onde a gente pausou, onde desistiu, o que reassistiu, a que horas, em que tela, o que abandonou no meio. &#201; uma das plataformas com o retrato mais detalhado do comportamento das suas clientes que existe. E a recomenda&#231;&#227;o delas &#233; uma das mais estudadas da ind&#250;stria.</p><p>Ent&#227;o por que a gente ainda entra, passa vinte minutos rolando a tela, e sai sem assistir nada?</p><p>Fiz essa pergunta para 159 pessoas, em conversas pelo WhatsApp. As respostas foram notavelmente parecidas, e apontam todas para o mesmo lugar.</p><h3>As pessoas saem da Netflix para decidir o que ver na Netflix</h3><p>Esse foi o padr&#227;o mais forte, e o mais revelador. Um monte de gente descreveu o mesmo ritual: abre a Netflix, n&#227;o decide, sai para outro lugar, e &#224;s vezes volta.</p><blockquote><p>&#8220;Primeiro entro na plataforma, confiro as sugest&#245;es e novidades, mas acabo decidindo com o ChatGPT ou com algum resultado de busca no Google.&#8221;</p><p>&#8220;Fico pesquisando reviews online, tipo IMDb. Coisa que a Netflix n&#227;o oferece.&#8221;</p><p>&#8220;N&#227;o vejo nenhum filme sem saber a nota da cr&#237;tica antes.&#8221;</p></blockquote><p>Pensa no que isso significa. Uma das plataformas com mais dados sobre o que a gente gosta n&#227;o consegue nos convencer a apertar o play sem que busquemos a opini&#227;o de um cr&#237;tico, de um amigo, ou de um estranho no TikTok. O dado n&#227;o fecha a decis&#227;o. A conversa com outra pessoa fecha.</p><h3>O dado sabe o que a gente viu. N&#227;o sabe quem est&#225; no sof&#225;.</h3><p>Uma resposta explicou, sozinha, uma das maiores limita&#231;&#245;es do dado de comportamento:</p><blockquote><p>&#8220;Uso a Netflix junto com meu namorado, ent&#227;o o algoritmo fica ampliado e me faz recomenda&#231;&#245;es erradas, mais direcionadas pra ele do que pra mim.&#8221;</p></blockquote><p>O hist&#243;rico &#233; de um perfil. Mas o perfil n&#227;o &#233; uma pessoa, &#233; &#224;s vezes duas, tr&#234;s, uma fam&#237;lia inteira num sof&#225;. O dado registra os cliques, n&#227;o sabe quem segura o controle hoje. E ningu&#233;m vai avisar o algoritmo &#8220;hoje sou s&#243; eu&#8221; ou &#8220;hoje &#233; o casal&#8221;. Isso s&#243; apareceria se algu&#233;m perguntasse. E foi o que aconteceu, usei a ReveLumi para perguntar e as pessoas responderam.</p><h3>&#8220;Depende do meu humor&#8221;</h3><p>Essa frase, ou uma vers&#227;o dela, apareceu v&#225;rias vezes:</p><blockquote><p>&#8220;Depende do dia. H&#225; dias que escolho qualquer coisa, h&#225; dias que navego mais.&#8221;</p><p>&#8220;Depende muito do meu humor.&#8221;</p></blockquote><p>O que a gente quer assistir numa ter&#231;a exausta depois do trabalho n&#227;o tem nada a ver com o que a gente quer numa sexta relaxada. &#201; a mesma pessoa, o mesmo hist&#243;rico, o mesmo dado. O que mudou foi o estado da pessoa naquela noite. E o estado de esp&#237;rito n&#227;o est&#225; no hist&#243;rico de cliques. Nenhuma quantidade de dados comportamentais captura o &#8220;como voc&#234; est&#225; se sentindo hoje&#8221;, porque isso n&#227;o &#233; um comportamento passado, &#233; um contexto presente. S&#243; uma pergunta alcan&#231;a.</p><h3>O custo de n&#227;o saber o porqu&#234;</h3><p>O mais interessante &#233; que a falha n&#227;o &#233; s&#243; de experi&#234;ncia. Ela vira problema de neg&#243;cio, na voz das pr&#243;prias pessoas:</p><blockquote><p>&#8220;Era para ser um momento de relaxamento e n&#227;o consigo relaxar. Muitas vezes acabo indo para outros streamings.&#8221;</p><p>&#8220;J&#225; aconteceu de desistir de assistir, mas o mais comum &#233; ir buscar algo em outro streaming.&#8221;</p></blockquote><p>Quando a recomenda&#231;&#227;o n&#227;o entende a pessoa, ela n&#227;o fica olhando para sempre. Ela sai. Vai para o concorrente ou desiste da noite. Todo o dado do mundo n&#227;o impede isso, porque o que falta n&#227;o &#233; mais dado sobre o passado dela. &#201; entender o que ela quer agora e por qu&#234;.</p><h3>O que isso tem a ver com o seu produto</h3><p>A Netflix &#233; o caso extremo e &#233; por isso que ele ensina. Se uma das empresas com mais dados comportamentais do planeta ainda perde a pessoa na hora de decidir, o que te faz achar que o seu dashboard responde a todas as perguntas do seu produto?</p><p>Dado &#233; um retrato do que aconteceu. Ele &#233; bom nisso e a gente precisa dele. Mas ele n&#227;o conta por que aconteceu, nem o que a pessoa sentia, nem o que ela quer agora. Esse peda&#231;o, o mais importante para decidir, mora num lugar s&#243;, na conversa.</p><p>Dados dizem o qu&#234;. Clientes dizem o porqu&#234;. A Netflix tem dado de sobra, e mesmo assim a gente sai para perguntar a um amigo o que assistir. Talvez seja hora de perguntar &#224;s suas clientes o que os seus dados n&#227;o contam.</p><p>Quer sentir na pele? Experimente <a href="https://revelumi.com/c/06VP6MA2">ser entrevistado pela ReveLumi</a> sobre como voc&#234; escolhe o que assistir na Netflix.</p><h3>Marque uma sess&#227;o de demo com os founders</h3><p>Se voc&#234; quer conhecer mais sobre a ReveLumi, <a href="https://calendar.google.com/calendar/appointments/schedules/AcZssZ1l4U38tikGBmiJWc8jkB2wkj20Ev5VlH94cY88GZOGWZ3ZKG4R9gSjj3EFt__ovsBGkUZnggcH">agende uma sess&#227;o de demo com o Anderson e comigo</a> para a gente te mostrar o que a ReveLumi faz hoje, o que estamos pensando para ela no futuro e para tirar todas as suas d&#250;vidas.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[Do people open up to an AI?]]></title><description><![CDATA[Someone doubted it while talking to our agent. It was the best answer I could have asked for.]]></description><link>https://blog.revelumi.com/p/do-people-open-up-to-an-ai</link><guid isPermaLink="false">https://blog.revelumi.com/p/do-people-open-up-to-an-ai</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 29 Jun 2026 10:36:27 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/sera-que-as-pessoas-se-abrem-com">Portuguese version</a>]</p><p>In the middle of one of the surveys I ran about how product people do discovery, a participant stopped to raise an interesting doubt. She works with AI, sees the use of AI as a natural move, and still wondered:</p><blockquote><p>&#8220;People still prefer to talk to humans. Opening up to a human, not everyone is comfortable with a contact that, however human it feels, is still not a person on the other side.&#8221;</p></blockquote><p>And she went further. She suggested the way to settle the doubt:</p><blockquote><p>&#8220;The ideal would be a test, to validate whether people really don&#8217;t open up to something that isn&#8217;t human.&#8221;</p></blockquote><p>She was right to ask for the test. And the test was happening while she spoke.</p><p>Look at what she did. To tell us that people don&#8217;t open up to an AI, she opened up to an AI. She told us about her company&#8217;s culture, her own doubts, what would or wouldn&#8217;t make her adopt a tool. A long paragraph, thoughtful, sincere, written to our agent, on WhatsApp. Her doubt already carried its own answer.</p><p>And it wasn&#8217;t only her.</p><h3>What 149 conversations showed</h3><p>So far we have 149 conversations about how product people do discovery, all with valuable insight, many in long, reflective answers, the kind no one writes when they are afraid to open up.</p><p>One person vented an internal defeat they would hardly share in a meeting room:</p><blockquote><p>&#8220;I always defended this, but I was torn apart.&#8221;</p></blockquote><p>Another acknowledged, without defensiveness, a limit of their own seniority:</p><blockquote><p>&#8220;I don&#8217;t interview any users right now. Since I&#8217;m junior, my discovery starts there.&#8221;</p></blockquote><p>Another exposed a failure of their own employer, with rare frankness:</p><blockquote><p>&#8220;At the bank where I work, no continuous discovery happens at all. We race against the clock and there is always rework.&#8221;</p></blockquote><p>No one is obliged to confess a defeat to a form. No one admits their own inexperience, or criticizes their own employer, in a cold survey. These people were not filling out a questionnaire. They were having a conversation.</p><h3>Why people open up</h3><p>What we see in the dynamic of conversations with AI is that three things lower people&#8217;s guard, and none of them depends on pretending the agent is human:</p><ul><li><p>The first is the person&#8217;s own time. The conversation is asynchronous. No one is waiting on the other side, no one is watching the clock. The person answers when they want, the way they want, with as much depth as they want. The pressure to perform that exists in a scheduled call, where you feel you are taking up someone&#8217;s time, is gone.</p></li><li><p>The second is the absence of judgment. Saying &#8220;I&#8217;m junior, my discovery starts there&#8221; to another professional costs something. Saying it to an agent that only wants to understand costs nothing. No audience, no hierarchy, no risk of looking small. The hesitation fades because there is no one to impress.</p></li><li><p>The third is the place. The conversation happens on WhatsApp, where the person already lives. It is not a strange link, not a video call room, not a new piece of software to learn. It is the same app where they talk to family and to their team. The setting is familiar, and the familiar relaxes us.</p></li></ul><h3>The doubt was right, and so was the answer</h3><p>The participant asked whether people open up to something that isn&#8217;t human. The honest answer is: yes, and sometimes more than they would to a human. Not because the agent fools anyone, no one there thinks they are talking to a person. But because the absence of another human removes exactly what holds a conversation back, the clock, the judgment, the formality.</p><p>She asked for a test. So far, 149 conversations, including the one with the person who had the doubt, have shown that people do open up to an AI.</p><p>Data tells us what happened. Customers tell us why. And, by all indications, they tell us why with a frankness that may be even greater than in conversations with humans.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[Será que as pessoas se abrem com uma IA?]]></title><description><![CDATA[Uma pessoa duvidou disso enquanto conversava com a nossa agente. Foi a melhor resposta que poder&#237;amos ter!]]></description><link>https://blog.revelumi.com/p/sera-que-as-pessoas-se-abrem-com</link><guid isPermaLink="false">https://blog.revelumi.com/p/sera-que-as-pessoas-se-abrem-com</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Sun, 28 Jun 2026 18:33:22 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>No meio de uma das pesquisas que rodei sobre como pessoas de produto fazem discovery, uma participante parou para levantar uma d&#250;vida interessante. Ela trabalha com IA, acha o uso de IA um movimento natural, e mesmo assim ponderou:</p><blockquote><p>&#8220;As pessoas ainda preferem falar com humanos. Se abrir com humano, n&#227;o &#233; todo mundo que v&#234; com bons olhos um contato que, por mais humanizado que seja, ainda assim n&#227;o &#233; uma pessoa ali do outro lado.&#8221;</p></blockquote><p>E foi al&#233;m. Sugeriu o caminho para resolver a d&#250;vida:</p><blockquote><p>&#8220;O ideal seria talvez um teste, para validar se as pessoas realmente n&#227;o se abrem com algo que n&#227;o &#233; humano.&#8221;</p></blockquote><p>Ela tinha raz&#227;o em pedir o teste. E o teste estava acontecendo enquanto ela falava.</p><p>Repare no que ela fez! Para nos dizer que as pessoas n&#227;o se abrem com uma IA, ela se abriu com uma IA. Contou da cultura da empresa dela, das d&#250;vidas dela, do que a faria ou n&#227;o adotar uma ferramenta. Um par&#225;grafo longo, pensado, sincero, escrito para a nossa agente, no WhatsApp. A d&#250;vida dela j&#225; vinha com a resposta embutida.</p><p>E n&#227;o foi s&#243; ela.</p><h3>O que 149 conversas mostraram</h3><p>At&#233; o momento temos 149 conversas sobre como pessoas de produto fazem discovery, todas com insights valiosos, muitas em respostas longas, reflexivas, das que n&#227;o se escreve com receio de se abrir.</p><p>Uma pessoa desabafou uma derrota interna que dificilmente contaria numa sala de reuni&#227;o:</p><blockquote><p>&#8220;Eu sempre defendi isso, mas fui dilacerado.&#8221;</p></blockquote><p>Outra reconheceu, sem defensiva, uma limita&#231;&#227;o da pr&#243;pria senioridade:</p><blockquote><p>&#8220;Atualmente eu n&#227;o entrevisto nenhum usu&#225;rio. Como sou j&#250;nior, meu discovery come&#231;a por a&#237;.&#8221;</p></blockquote><p>Outra exp&#244;s a falha do pr&#243;prio empregador, com uma franqueza rara:</p><blockquote><p>&#8220;No banco em que trabalho, n&#227;o &#233; realizado nenhum tipo de discovery cont&#237;nuo. Corremos contra o tempo e sempre temos retrabalho.&#8221;</p></blockquote><p>Ningu&#233;m &#233; obrigado a confessar uma derrota para um formul&#225;rio. Ningu&#233;m admite a pr&#243;pria inexperi&#234;ncia ou critica o pr&#243;prio empregador numa pesquisa fria. Essas pessoas n&#227;o estavam preenchendo um question&#225;rio. Estavam conversando.</p><h3>Por que as pessoas se abrem</h3><p>O que vemos na din&#226;mica de conversas com IA &#233; que tr&#234;s coisas baixam a guarda das pessoas, e nenhuma delas depende de fingir que a agente &#233; humana:</p><ul><li><p>A primeira &#233; o tempo da pessoa. A conversa &#233; ass&#237;ncrona. Ningu&#233;m est&#225; esperando do outro lado; ningu&#233;m olha no rel&#243;gio. A pessoa responde quando quer, do jeito que quer, com a profundidade que quiser. Some a press&#227;o de performar que existe numa call agendada, onde voc&#234; sente que est&#225; tomando o tempo de algu&#233;m.</p></li><li><p>A segunda &#233; a aus&#234;ncia de julgamento. Dizer &#8220;sou j&#250;nior, meu discovery come&#231;a por a&#237;&#8221; para outro profissional custa. Dizer para uma agente que s&#243; quer entender, n&#227;o custa. Sem plateia, sem hierarquia, sem o risco de parecer pouco. O receio diminui porque n&#227;o h&#225; ningu&#233;m para impressionar.</p></li><li><p>A terceira &#233; o lugar. A conversa acontece no WhatsApp, onde a pessoa j&#225; vive. N&#227;o &#233; um link estranho, n&#227;o &#233; uma sala de videoconfer&#234;ncia, n&#227;o &#233; um software novo para aprender. &#201; o mesmo app onde ela fala com a fam&#237;lia e com o time. O ambiente &#233; familiar, e o familiar relaxa.</p></li></ul><h3>A d&#250;vida estava certa, e a resposta tamb&#233;m</h3><p>A participante perguntou se as pessoas se abrem com algo que n&#227;o &#233; humano. A resposta honesta &#233;: sim, e &#224;s vezes mais do que se abririam com um humano. N&#227;o porque a agente engana algu&#233;m; ningu&#233;m ali pensa que est&#225; falando com uma pessoa. Mas porque a aus&#234;ncia de outro humano remove exatamente o que trava a conversa, o rel&#243;gio, o julgamento, a formalidade.</p><p>Ela pediu um teste. At&#233; o momento, 149 conversas, incluindo a da pessoa que tinha a d&#250;vida, mostraram que as pessoas se abrem sim com uma IA.</p><p>Dados dizem o qu&#234;. Clientes dizem o porqu&#234;. E, ao que tudo indica, dizem o porqu&#234; com uma franqueza que pode at&#233; ser maior do que em conversas com humanos.<br></p>]]></content:encoded></item><item><title><![CDATA[The ROI of talking to customers]]></title><description><![CDATA[82 conversations in two days, run by the tool itself, and what they revealed about who decides to invest in listening to customers.]]></description><link>https://blog.revelumi.com/p/the-roi-of-talking-to-customers</link><guid isPermaLink="false">https://blog.revelumi.com/p/the-roi-of-talking-to-customers</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Mon, 22 Jun 2026 11:18:25 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>[<a href="https://listentodecide.substack.com/p/o-roi-de-conversar-com-clientes">Portuguese version</a>]</p><p>A month after launching ReveLumi, I started noticing a pattern in the conversations I had while presenting the product. The people who feel the pain of not talking to customers are rarely the ones who hold the budget to fix it. I decided to use ReveLumi, our research agent on WhatsApp, to listen to more people and test that hypothesis.</p><p>I created a survey, generated a link, and posted it in a few product communities, asking for help with a discovery about how people do discovery. A meta-discovery. The first message went out at 11am. Three hours later, I had 42 conversations with insight. Two days later, 82, with 46 completed. I scheduled nothing, recruited no one, sat in on not a single call. The synthesis was waiting for me when I opened the report.</p><p>The finding: the biggest barrier to talking to customers is operational, scheduling and recruiting. And the way I ran this research removed that barrier entirely. The method is the proof of the finding.</p><h3>The pain is operational, and it is the same everywhere</h3><p>People described the same difficulty over and over. Calendars that never line up. Recruiting that drags. People who do not show up. Emails that go unanswered.</p><blockquote><p>&#8220;Plenty of willingness, but it is extremely hard to make it happen without all the scheduling work: booking, canceling, no-shows, showing up with no time.&#8221;</p></blockquote><p>This was the majority view, by a wide margin. People know they should talk to users more. They just cannot, because operations eat the time.</p><blockquote><p>&#8220;We talk to users less than we would like. Too many urgent fronts, more technical work than discovery itself.&#8221;</p></blockquote><p>At a bank, someone was blunt about the cost of this:</p><blockquote><p>&#8220;We do no continuous discovery. We interview users only when we are about to start building. It is rushed, and there is always rework.&#8221;</p></blockquote><h3>The budget decision lives far from the pain</h3><p>This is the finding I suspected but had not confirmed. The people who feel the pain are in product and design. The people who approve the budget are above them.</p><blockquote><p>&#8220;The demand starts with product and design, the ROI has to be very well proven, and then a business case goes up that the CTO has to approve, as long as there is room in the annual budget.&#8221;</p></blockquote><p>A product participant captured the misalignment in one line:</p><blockquote><p>&#8220;Only product and design talk to customers. So only we feel this need.&#8221;</p></blockquote><p>And another revealed the quietest consequence of it. When the person who feels the pain sees no solution, the pain never even reaches leadership:</p><blockquote><p>&#8220;I do not speak for leadership. I do not take it to them myself, because there is no solution today.&#8221;</p></blockquote><p>The person who feels the pain is not the person who decides the purchase. And sometimes does not even carry the pain forward.</p><h3>Removing the barrier pays off in better decisions</h3><p>When we asked what would change with continuous, direct access to users, the answers converged. Faster decisions, less noise, less rework.</p><blockquote><p>&#8220;We would have faster decisions with less noise. Today the need reaches us already interpreted by other areas, and context gets lost along the way.&#8221;</p></blockquote><p>Another summed up the gain in confidence:</p><blockquote><p>&#8220;It would help validate hypotheses faster, reduce rework, and increase confidence in prioritization.&#8221;</p></blockquote><p>The reward for solving an operational problem turns out to be strategic. And it can be measured.</p><h3>The ROI, in two parts</h3><p>The return on improving discovery has two parts, of different natures.</p><p><strong>Operational ROI:</strong></p><ul><li><p>Running 82 qualitative conversations by hand (recruiting, scheduling, conducting, transcribing, analyzing, synthesizing) would take weeks of work.</p></li><li><p>I had the full synthesis ready in two days, with the first results in three hours, scheduling nothing.</p></li></ul><p><strong>Strategic ROI:</strong></p><ul><li><p>Weeks of development saved for every bad hypothesis the team does not chase. And most hypotheses are bad. Ron Kohavi, who led experimentation at Microsoft and Airbnb, documented that about two thirds of product ideas fail to improve the metric they were meant to improve, and that in mature companies this rate reaches 80 or 90%.</p></li><li><p>The cost of keeping the team chasing the wrong thing. A squad of six people costs around 200 thousand reais per month in Brazil, 80 thousand dollars in the US, 50 thousand euros in Europe. A month in the wrong direction is that entire amount burned.</p></li><li><p>The cost avoided of building the wrong product. In 2010, at Locaweb, a conversation with customers led us to abandon a marketplace before writing a single line of code. The data said go ahead, the conversation revealed why not. We avoided the cost of building the wrong thing, and also of operating a product doomed from conception.</p></li></ul><p>Using a tool to help talk to customers does not just save operational hours. It saves the cost of a whole squad running in the wrong direction, and sometimes prevents the mistake you would only notice too late.</p><h3>What we are doing with this</h3><p>The second finding is uncomfortable for us. It means the people who love ReveLumi most, the product manager and the researcher drowning in scheduling, are often not the ones who can buy it.</p><p>There are two ways to react. Try to create the pain in those who hold the budget, which is slow and expensive. Or arm the people who already feel the pain with what they need to make the case. We are choosing the second. The product manager and the researcher who want to use ReveLumi should not have to build the business case alone. That becomes our job, to hand them the ROI, the metrics, the argument a director will approve.</p><p>These insights did not come from a strategy offsite or a consulting deck. They came from 82 conversations that, two years ago, would have taken weeks and probably never have happened. The obstacles would have won.</p><p>Data tells us what happened. Customers tell us why. The reason teams stop at the data is that the conversation is too hard to have. Remove that friction, and the why comes back. That is what we are building. I had a real question about ReveLumi itself, and I cleared it up in a few hours, with dozens of conversations with real users.</p><h3>To take to leadership</h3><p>If you read this far, you probably feel this operational pain yourself, and the decision to invest is probably not only yours. So use this text. Forward it to whoever decides the budget at your company. The numbers are here, the logic is here, the ROI is here. Talking to customers leads to faster decisions and less rework, and there is now a way to have those conversations without the friction that always got in the way. That is the part the person who approves the investment needs to hear.</p><p></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.revelumi.com/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Listen to Decide! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div>]]></content:encoded></item><item><title><![CDATA[O ROI de conversar com clientes]]></title><description><![CDATA[82 conversas em dois dias, conduzidas pela pr&#243;pria ferramenta, e o que elas revelaram.]]></description><link>https://blog.revelumi.com/p/o-roi-de-conversar-com-clientes</link><guid isPermaLink="false">https://blog.revelumi.com/p/o-roi-de-conversar-com-clientes</guid><dc:creator><![CDATA[Joca Torres]]></dc:creator><pubDate>Sun, 21 Jun 2026 10:42:37 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!q5bw!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd7b39fc8-e9d8-4a3c-8566-1f2369b16b7d_512x512.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Ap&#243;s <a href="https://www.youtube.com/watch?v=MfQCy7SFMNc&amp;cc_lang_pref=en&amp;cc_load_policy=1&amp;t=103s">lan&#231;ar a ReveLumi h&#225; um m&#234;s</a>, comecei a notar um padr&#227;o em algumas conversas que tive ao apresentar o produto: quem sente a dor de n&#227;o falar com clientes raramente &#233; quem tem o budget para resolver. Decidi usar a ReveLumi, nossa agente de pesquisa via WhatsApp, para ouvir mais pessoas e (in)validar essa hip&#243;tese.</p><p>Criei uma pesquisa, gerei um link e o postei em alguns grupos de produto pedindo ajuda num discovery sobre como as pessoas fazem discovery. Um meta-discovery. A primeira mensagem saiu &#224;s 11h. Tr&#234;s horas depois, eu j&#225; tinha 42 conversas com insight. Dois dias depois, 82, com 46 completas. N&#227;o agendei nada, n&#227;o recrutei ningu&#233;m, n&#227;o participei de uma &#250;nica conversa. A s&#237;ntese estava pronta quando abri o relat&#243;rio.</p><p>O resultado da pesquisa &#233; que a maior barreira para falar com cliente &#233; operacional, agenda e recrutamento. E a forma como rodei essa pesquisa removeu essa barreira por completo. O m&#233;todo &#233; a prova do achado.</p><h3>A dor &#233; operacional, e &#233; a mesma em todo lugar</h3><p>As pessoas descreveram a mesma dificuldade de forma recorrente. Agendas que n&#227;o batem. Recrutamento que se arrasta. Gente que n&#227;o aparece. E-mails sem resposta.</p><blockquote><p>&#8220;Muita vontade, mas &#233; extremamente dif&#237;cil isso acontecer sem trabalho de agendamento: agendar, cancelar, n&#227;o aparecer, aparecer sem tempo.&#8221;</p></blockquote><p>Essa foi a vis&#227;o da maioria, com folga. As pessoas sabem que deveriam falar mais com usu&#225;rias. S&#243; n&#227;o conseguem porque a opera&#231;&#227;o engole o tempo.</p><blockquote><p>&#8220;Falamos menos do que gostar&#237;amos com as usu&#225;rias. Muitas frentes com urg&#234;ncia, trabalho mais t&#233;cnico do que discovery propriamente dito.&#8221;</p></blockquote><p>Em um banco, algu&#233;m foi direta sobre o custo disso:</p><blockquote><p>&#8220;N&#227;o fazemos nenhum discovery cont&#237;nuo. Entrevistamos as usu&#225;rias quando estamos prestes a iniciar o desenvolvimento. &#201; corrido e sempre temos retrabalho.&#8221;</p></blockquote><h3>A decis&#227;o de budget para resolver mora longe da dor</h3><p>Esse &#233; o achado que eu suspeitava e n&#227;o tinha confirmado. Quem sente a dor est&#225; em produto e design. Quem aprova o budget est&#225; acima.</p><blockquote><p>&#8220;A demanda parte de produto e design, o ROI precisa estar muito bem provado, e da&#237; &#233; feito um business case que precisa ser aprovado pelo CTO, mediante ter espa&#231;o no or&#231;amento anual.&#8221;</p></blockquote><p>Uma participante de produto cravou o desalinhamento em uma frase:</p><blockquote><p>&#8220;S&#243; produto e design falam com clientes. Ent&#227;o s&#243; n&#243;s sentimos essa necessidade.&#8221;</p></blockquote><p>E outra mostrou a consequ&#234;ncia mais silenciosa disso. Quando quem sente a dor n&#227;o enxerga uma solu&#231;&#227;o, nem chega a levar o problema para a lideran&#231;a:</p><blockquote><p>&#8220;N&#227;o falo pela lideran&#231;a. Eu mesma n&#227;o levo para eles, porque n&#227;o tem solu&#231;&#227;o hoje.&#8221;</p></blockquote><p>Quem sente a dor n&#227;o &#233; quem decide a compra. E, &#224;s vezes, nem leva a dor adiante.</p><h3>Remover a barreira se paga em decis&#227;o melhor</h3><p>Quando perguntamos o que mudaria com acesso cont&#237;nuo e direto &#224;s usu&#225;rias, as respostas convergiram. Decis&#227;o mais r&#225;pida, menos ru&#237;do, menos retrabalho.</p><blockquote><p>&#8220;Ter&#237;amos decis&#245;es mais r&#225;pidas e com menos ru&#237;do. Hoje a necessidade chega j&#225; interpretada por outras &#225;reas, e o contexto se perde no caminho.&#8221;</p></blockquote><p>Outra resumiu o ganho de confian&#231;a:</p><blockquote><p>&#8220;Ajudaria a validar hip&#243;teses com mais velocidade, reduzir retrabalho e aumentar a confian&#231;a nas prioriza&#231;&#245;es.&#8221;</p></blockquote><p>O pr&#234;mio por resolver um problema operacional acaba sendo estrat&#233;gico.</p><h3>O ROI, em duas partes</h3><p>O retorno do investimento em melhorar o discovery tem duas partes, de naturezas diferentes.</p><p><strong>ROI operacional:</strong></p><ul><li><p>Fazer 82 conversas qualitativas &#224; m&#227;o (recrutar, agendar, conduzir, transcrever, analisar, sintetizar) levaria semanas de trabalho.</p></li><li><p>Eu tive a s&#237;ntese completa pronta em dois dias, sem agendar nada, com os primeiros resultados em 3 horas.</p></li></ul><p><strong>ROI estrat&#233;gico:</strong></p><ul><li><p>Semanas de desenvolvimento economizadas a cada hip&#243;tese ruim que o time deixa de perseguir. E a maioria das hip&#243;teses &#233; ruim. Ron Kohavi, que liderou experimenta&#231;&#227;o na Microsoft e na Airbnb, documentou que cerca de dois ter&#231;os das ideias de produto n&#227;o melhoram a m&#233;trica que deveriam e que, em empresas maduras, essa taxa chega a 80% ou 90%.</p></li><li><p>O custo de manter o time perseguindo a coisa errada. Uma squad de seis pessoas custa, no Brasil, por volta de 200 mil reais por m&#234;s; nos EUA, 80 mil d&#243;lares; e na Europa, 50 mil euros. Um m&#234;s na dire&#231;&#227;o errada &#233; esse valor inteiro queimado.</p></li><li><p>O custo evitado de construir o produto errado. Em 2010, na Locaweb, uma conversa com clientes nos fez <a href="https://listentodecide.substack.com/p/dados-dizem-o-que-clientes-dizem">abandonar o desenvolvimento de um marketplace</a> antes de escrever qualquer linha de c&#243;digo. Os dados diziam para seguir; a conversa revelou por que n&#227;o. Evitamos o custo de construir a coisa errada e tamb&#233;m o de operar um produto fadado ao fracasso desde a concep&#231;&#227;o.</p></li></ul><p>Usar uma ferramenta para ajudar a conversar com clientes n&#227;o economiza s&#243; horas de trabalho operacional. Economiza o custo de uma squad inteira rodando na dire&#231;&#227;o errada e, &#224;s vezes, evita o erro que voc&#234; s&#243; perceberia tarde demais.</p><h3>O que estamos fazendo com isso</h3><p>O segundo achado &#233; desconfort&#225;vel para n&#243;s. Significa que a pessoa que mais ama a ReveLumi, a gerente de produto e a researcher afogadas em agendamento, muitas vezes n&#227;o s&#227;o quem pode comprar.</p><p>D&#225; para reagir a isso de dois jeitos. Tentar criar a dor em quem tem o budget, o que &#233; lento e caro. Ou armar quem j&#225; sente a dor com o que ela precisa para defender a compra. Nesse momento escolhemos o segundo. A gerente de produto e a researcher que querem usar a ReveLumi n&#227;o precisam montar o business case sozinhas. Esse vira o nosso trabalho: entregar a ela o ROI, as m&#233;tricas, o argumento que uma diretora aprova.</p><p>Esses insights n&#227;o vieram de um offsite de estrat&#233;gia ou de um slide de consultoria. Veio de 82 conversas que, dois anos atr&#225;s, levariam semanas e provavelmente nunca teriam acontecido. Os obst&#225;culos teriam vencido.</p><p>Dados dizem o qu&#234;. Clientes dizem o porqu&#234;. A raz&#227;o dos times pararem no dado &#233; que a conversa &#233; dif&#237;cil demais de ter. Remova esse atrito e o porqu&#234; volta. &#201; isso que estamos construindo. Eu tinha uma d&#250;vida real sobre a pr&#243;pria ReveLumi e clareei em poucas horas, com dezenas de conversas com usu&#225;rias reais.</p><h3>Para levar &#224; lideran&#231;a</h3><p>Se voc&#234; leu at&#233; aqui, &#233; prov&#225;vel que sinta essa dor operacional na pele, e que a decis&#227;o de investir n&#227;o seja s&#243; sua. Ent&#227;o use este texto. Encaminhe para quem decide o budget na sua empresa. Os n&#250;meros est&#227;o aqui, a l&#243;gica est&#225; aqui, o ROI est&#225; aqui. A conversa com cliente gera decis&#227;o mais r&#225;pida e menos retrabalho, e hoje existe uma forma de ter essas conversas sem o atrito que sempre as impediu. Essa &#233; a parte que cabe a quem aprova o investimento ouvir.</p>]]></content:encoded></item></channel></rss>