We were making product decisions the way most teams do: by whoever spoke loudest in the last meeting. We called it “stakeholder alignment.” Looking back, it was just organized guessing.
Then we tried something different.
How Product Decisions Actually Get Made
Here’s the honest version of how roadmaps get built at most companies. The PM remembers a complaint from a user call three months ago. A senior stakeholder has a strong intuition from their last customer visit. Someone in leadership read an article about a competitor feature. All of this gets synthesized in a planning meeting, under time pressure, into a list of things to build.
Nobody is lying. Nobody is being reckless. But the distance between “what we think users need” and “what users actually need” quietly grows. Assumptions calcify. Features ship. Results are mixed. The cycle repeats.
At Voxy — an AI-powered language learning platform for corporate learners — I recognized this pattern and decided to break it. Not with a big process overhaul. With a weekly habit.
The Teresa Torres Method
I’d been following Teresa Torres’s work on Continuous Discovery Habits for a while. The framework is deceptively simple: talk to at least one user every week, every week, without exception. Build an opportunity solution tree so the whole team can see why you’re building what you’re building. Test your assumptions before you bet a sprint on them.
What makes it work isn’t any single element — it’s the compounding effect. One user interview is anecdote. Twelve consecutive weeks of interviews, with structured notes, with a shared repository, with patterns emerging across conversations — that’s something you can actually make decisions from.
Think of it like weekly therapy for your product assumptions. Without it, your assumptions calcify. They start feeling like facts. With it, they stay flexible. You hold them more lightly because you’re in a constant low-stakes process of checking them.
What Implementation Actually Looked Like
The critical design choice was making the cadence non-optional. Not “we’ll do interviews when we have bandwidth.” Dedicated calendar slots, recurring, protected. PM and designer attended together — not because we were being precious about it, but because when both of you hear the same user say the same thing, there’s no filter between insight and decision.
Participants were recruited in-app, which kept the sample fresh and avoided the usual problem of only ever talking to your most engaged (and unrepresentative) power users. Notes went into a shared repository immediately after each session — not polished, just captured — so the whole squad could see what we were learning in real time.
The opportunity solution tree was the tool that changed team dynamics more than anything else. Suddenly, engineers and designers could see why we were building something, not just what. The tree made the reasoning visible. It turned “trust me, the PM thinks this is important” into “here’s the user need, here’s the opportunity, here’s why this solution addresses it.”
🚧 Need more context: What was the specific product area where continuous discovery was first piloted at Voxy? What was the team composition — how many people were involved in the weekly interview cadence? Can you share a specific user insight that led directly to a concrete product decision?
What Changed
The discovery-to-decision cycle was cut by 50%. That number deserves some unpacking, because it sounds like an efficiency metric but it’s really a quality metric. We weren’t moving faster because we were cutting corners. We were moving faster because we had less rework. When you’ve already stress-tested your assumptions against real users, you spend less time in planning debates about what users want. You know. You build. You ship.
The other change was harder to measure but more important: the team stopped feeling like we were guessing. There’s a particular kind of confidence that comes from having talked to seven users in the past seven weeks. It’s not certainty — you still don’t know everything — but it’s the right kind of informed humility.
“Continuous discovery doesn’t eliminate uncertainty. It makes uncertainty productive — something you actively work through rather than something that paralyzes you in planning.”
The Product Talk Recognition
The work was featured in a Product Talk article — Teresa Torres’s own publication — as a model for adopting structured discovery in a scaling team. That recognition mattered less as a credential and more as a signal: the approach was replicable enough that it was worth writing about for other PMs trying to do the same thing.
What I’d Do Differently
I’d start with the opportunity solution tree earlier — before the first interview, not after a few weeks of them. Framing the opportunity space upfront gives the interviews sharper focus. Without it, early sessions can be exploratory to the point of being hard to synthesize.
I’d also be more deliberate about closing the loop with the users we interviewed. We got good at collecting insight. We weren’t always good at coming back and saying “here’s what we built because of what you told us.” That’s a missed relationship-building opportunity, and it would have made recruitment for future rounds easier too.