--- title: How We Talk to Early Testers (And Why Most Startups Do It Wrong) description: We talk to users every single day. Not through surveys or dashboards — actual conversations. Here is why that matters and how we do it at scale. canonical: https://nowah.xyz/blog/how-we-talk-to-users lastModified: "2026-08-07T08:32:23.742Z" --- # How We Talk to Early Testers (And Why Most Startups Do It Wrong) We talk to users every single day. Not through surveys or dashboards — actual conversations. Here is why that matters and how we do it at scale. There is a difference between data and understanding. Analytics dashboards tell you what users do. Conversations tell you why. Most startups default to analytics. They track clicks, measure funnels, and A/B test button colors. These are useful activities. But they miss the [most important](/blog/why-speed-is-most-important-feature) signal: why someone chose what they chose, what frustrated them, and what they actually needed versus what they asked for. We talk to early testers every day. Not quarterly research sprints. Daily conversations. ## The difference between what users say and what they do ![Illustration for this section](https://pics.nowah.xyz/website-media/founder-014-img-1.webp) The most important thing [user research](/blog/user-research-ai-products-what-learned) has taught us is that people are unreliable narrators of their own preferences. Ask a traveler what they want and they will say "[cheap flights](/blog/cheap-flights-2026-ai-finds-deals)." Watch what they actually choose when presented with options and the story is different. They choose the direct flight that costs a hundred dollars more. They pick the hotel that is walkable to restaurants, not the one that saves thirty dollars a night. This gap between stated preferences and revealed behavior is enormous in travel. And you can only see it through conversation, not surveys. When we talk to early testers, we ask open-ended questions. "Tell me about the last trip you planned." "What was the hardest part?" "What made you decide on that hotel?" The answers are rich with context that no dashboard can provide. ## Our process: daily, not quarterly We do not batch user research into quarterly sprints. We bake it into daily operations. Every piece of user feedback, whether it comes from a conversation, a support message, or a behavioral pattern in the data, goes into a running insight log. Once a week, we review the log and identify patterns. Patterns that repeat across multiple users get flagged for deeper investigation. Before any feature enters the engineering pipeline, it requires validation from at least ten user conversations. Not "ten users clicked a survey link." Ten actual conversations where someone described the problem, reacted to a proposed solution, and gave honest feedback. ## How user conversations shape our roadmap ![Supporting diagram](https://pics.nowah.xyz/website-media/founder-014-img-2.webp) Three specific examples of how direct user conversations changed what we built. First, streaming. Early private testers told us they felt anxious during the pause between sending a message and receiving a response. Analytics showed the average [response time](/blog/ten-second-rule-ai-response-time) was under ten seconds. That should have been fine. But the conversations revealed that the issue was not speed but visibility. Users did not know what was happening. Adding [streaming responses](/blog/streaming-ai-responses-real-time-chat), where users see the agent working in real time, came directly from these conversations. Second, post-booking. We were focused entirely on the search-to-book flow. Users kept asking: "What about after I book?" This question came up in enough conversations to make post-booking management a top priority, weeks before any analytics data would have surfaced the same insight. Third, voice input. We noticed in conversations that users described trips verbally with far more richness than they typed. One user said: "If I could just talk to it like I am talking to you right now, that would be perfect." Voice input was not on our roadmap. It is now a core feature. ## The most surprising thing The most surprising feedback we have received is how personal travel decisions actually are. We expected users to optimize for price and convenience. What we discovered is that travel decisions are deeply emotional. People choose destinations because of a memory, a movie, a friend's Instagram post. They avoid certain airports because of a bad experience years ago. They prefer specific hotel neighborhoods because of a feeling, not a data point. This insight fundamentally changed how we design the agent's conversational style. It is not a calculator. It is a travel companion that understands that "I want to go back to that feeling I had in Barcelona" is a valid search query. ## Start today If you are building a product and you are not talking to users daily, start this week. Schedule five calls. Ask open-ended questions. Listen more than you talk. And resist the urge to defend your product decisions during the conversation. The gap between what you think users want and what they actually need is always larger than you expect. Conversations close that gap. Dashboards do not. --- Nowah is an AI travel agent that searches and books real flights and hotels through conversation — no filters, no thirty open tabs. [Plan your next trip](https://app.nowah.xyz).