If you've been building for a while, you know the hardest part isn't shipping features — it's knowing what's actually worth building. Most early-stage roadmaps are packed with assumptions: some ideas feel exciting but won't grow the business, others have potential only if you shape them right. After 15 years helping early and growth-stage startups fix UX bottlenecks, we keep seeing the same thing: the moment you bring real user feedback into the process, the endless internal debates stop. People move faster, decisions get clearer, and the roadmap finally makes sense. Here's how we run it.

Start with what you want to learn

Before talking to anyone, get clear on why. Are you testing a new flow, watching what people actually do, discovering pain points, or understanding why customers pick you over competitors? Different questions, different interview styles, sometimes different people. Skip this step and the interview feels messy and useless. Then prep three to five talking points — not a word-for-word script, which kills the conversation by keeping you focused on the next question instead of listening. Be present and ask follow-ups: "tell me more about this," "talk me through that moment."

A few structures we use: the Mom Test's five questions are a simple place to start, anchored in the past ("tell me about the last time you did X — what was the hardest part?"), because asking about future behaviour just gets people making things up. For customers who picked you or a competitor, dig into how they ended up choosing — the trigger, where and when they found the solution, what else they considered. For past clients who moved on, understand how and why. For open discovery, find out whether people have the problem, whether they're aware of it, and how badly they want it solved.

Recruit, run, and analyse

Recruiting doesn't need to be fancy. If you have customers, email them or add an in-product popup, always with an effortless scheduling link and reminders a day before and an hour or two before. No customers yet? Start with your network (but know that people close to you may soften the truth), search LinkedIn, join niche communities, run small ads, ask for intros. You only need a handful to get signal.

When you run sessions, come prepared — a working prototype if it's a usability test, a way to take notes, and ideally a teammate so one person can focus on the conversation while the other captures follow-ups. Be curious, not convincing: you're there to learn, not to sell. Keep discovery and sales calls separate — that line gets crossed fast and goes south quickly. Record when you can; you'll catch things later you missed live, and you can pull real quotes. Afterward, sit down and analyse properly — don't just jot down what you remember, which costs you depth and nuance. We use Dovetail for tagging and highlights and Airtable to log every piece of feedback and tie it to specific parts of the product. That's what shapes the roadmap — not opinions and half-remembered feedback.

We're not big fans of unmoderated testing, but when you're in a rush it can work for simple prototypes — uploading a flow and seeing whether people get from A to B, or sending a screenshot and asking where they'd start a task to build a heat map. But if you can run live interviews, always choose them first: they give you the nuance, emotion, hesitation, and frustration tools can't capture yet. User interviews and usability tests don't have to be a giant research project. They're about making better decisions faster — before you waste time building things nobody needs.

Want a second pair of eyes on your roadmap and what's actually worth building? Get in touch.