I built a roadmap from data, then one user interview changed everything

I spent a week analyzing data. 350 users, 50 daily actives, 67-second sessions. Built a report with features and go-to-market plans for the next month.

Then I talked to one user for 30 minutes and realized half of it was wrong.

The Setup

Building a web3 events app solo—Insidr, a Telegram mini app for India's web3 ecosystem. No download, no signup, just open and use.

The data was clear: 0.6% tried networking, 5.7% marked events. Short sessions, low engagement.

My diagnosis: passive experience, nothing to engage with.

My solution: Weekly polls, content reactions, live social proof, reputation systems. Goal: push sessions from 67 seconds to 3-5 minutes, make people check daily instead of weekly.

Felt solid. Data-driven. Ready to ship.

The Interview

Shrishail: recent grad, hackathon winner, working at a web3 startup. Sent 9 connection requests on the app.

Me: "What would make you use this more?"

Him: "Put a Quest banner at the top—it's buried. Add hackathon details in event cards. A job board would boost engagement."

Wait. We have Quests. He's asking for a banner because... he couldn't find it?

Him: "The networking feature is great—I can see LinkedIn and Twitter, connect before IBW."

Hold on. Only 0.6% of users tried networking. I assumed nobody wanted it. But he loves it. He's literally one of the 2 people using it.

Him: "I check weekly for new events."

Weekly. Not daily.

My entire roadmap assumed 67-second sessions were bad and weekly check-ins were a problem to solve.

What if weekly is... correct?

What I Got Wrong

I diagnosed from symptoms, not root causes.

• 67 seconds = bad engagement → Maybe it's perfect for a weekly event check?

• 0.6% networking = nobody wants it → Maybe they just can't find it?

• Low feature usage = need more features → Maybe just surface existing ones better?

I built solutions before understanding problems.

Polls about TON prices? That came from me. Content reactions? Saw it work elsewhere. None of this came from "here's a problem users face."

What I Did Next

Stopped building. Started watching.

Spent 3 hours on session recordings. Not skimming—actually watching users.

Found:

• Tabs looked terrible

• Info section blocked content

• Profile popup full of white space

• Can't see sent connection requests

Not feature gaps. Basic usability problems making the app feel half-baked.

Called the admin dashboard user. Listened for 15 minutes. Rebuilt it in 2 hours so he could actually use it.

Fixed the obvious stuff: Better tabs, collapsed info section, fixed profile layout, added "Sent Requests" view.

Nothing groundbreaking. Just made the app 30% more useable.

Same thing happened when I changed the homepage weeks ago—switched from IBW landing page to events listing. Session duration doubled overnight. 33 to 67 seconds.

Not from engagement hooks. From removing friction to what people actually wanted.

What Changed

Interviews before roadmaps. Doing 10 more over the next 4 days. Looking for patterns, not feature requests.

If 8/10 say "I check weekly," my 5-minute session goal is wrong.

If 8/10 say "didn't know networking existed," it's discovery, not features.

Fix basics before adding features. Session recordings show confusion that analytics never will.

Question the metrics. "Increase session duration" is meaningless without context. What job is this app doing? If it's "weekly event check," 67 seconds might be perfect.

Validate problems, not solutions. When someone says "add job board," ask: "Why? What's broken about current platforms?" Dig until you hit actual pain.

The Point

I'm a tech person. I know how to build things.

Product is different. It's not about building well—it's about building the right thing.

Data tells you what's happening. Only users tell you why.

This was my first user interview. Two weeks into thinking about product. Already caught myself optimizing for the wrong thing.

10 more interviews scheduled. Roadmap on hold until I find patterns. And I'm learning that good product work isn't about having answers—it's about asking better questions.

Building in public. Try the app if you're in India's web3 space—it's a Telegram mini app, no download or signup needed. Would love your feedback.