There’s a version of product management where decisions are made in quarterly planning rooms by people who haven’t spoken to a user in six months. I’ve worked in that version. It produces confident-sounding roadmaps and baffling outcomes.
Then I read Continuous Discovery Habits by Teresa Torres, and something clicked.
The premise is deceptively simple: talk to users every week, without exception. Not in a big formal research project. Not in a product review. Just a weekly interview. One conversation, every week, with someone who uses (or should use) your product.
That’s it. That’s the core of the framework.
What actually happens when you do it is anything but simple.
What “Continuous” Actually Means
Most product teams research in sprints. There’s a discovery phase, a synthesis phase, a validation phase — and then you build for six months and surface again to ask why users aren’t engaging.
Continuous discovery breaks this cycle. The insight doesn’t accumulate in a quarterly research report that everyone reads once and forgets. It flows in weekly, shaping how you interpret data, prioritize opportunities, and write specs.
At Voxy, we started with a single weekly interview cadence across the product trio — PM, designer, engineer — all listening together. Within four weeks, we had more real signal than from the previous two quarters of surveys combined. Not because surveys are useless, but because a person explaining why they abandoned a lesson tells you things a 1-5 rating scale simply cannot.
The Opportunity Solution Tree: A Map for Staying Honest
The part of continuous discovery that most teams skip is the Opportunity Solution Tree. It sounds like consulting jargon. It’s actually the thing that keeps you from fooling yourself.
The idea: your product exists to move a business outcome. Below that outcome sit opportunities — jobs users are trying to do, unmet needs, friction points. Below each opportunity sit possible solutions. Below each solution sit assumptions — the things that must be true for the solution to work.
Drawing this tree forces a discipline I didn’t know I was missing: you can only justify building something if you can trace it from a solution back up to an opportunity and back further to an outcome. If you can’t draw the line, you don’t build it.
At Voxy, this meant some very popular feature requests never made it into the tree. They were real requests from real users, but they didn’t connect to any opportunity that mattered for retention or language acquisition progress. Without the tree, we might have built them — and watched our velocity drop with nothing to show for it.
What Changes After Three Months
The discovery-to-decision cycle at Voxy went from roughly eight weeks to three. Here’s why:
Decisions stop being debates. When the team has talked to 12 users in the past month, there’s less room for gut-feel arguments. Someone says “users won’t understand the onboarding” and someone else says “actually, we talked to Claudia last Tuesday and she walked right through it.” The weekly cadence creates a shared knowledge base that dissolves opinion fights.
You stop second-guessing the roadmap. The quarterly planning exercise becomes less fraught. You’ve been collecting evidence. You know what problems are real. You know which ones are worsening. The roadmap writes itself.
Engineers get better requirements. When specs are grounded in specific user quotes, engineers ask better questions. “Is this about the same problem Maria described with the audio playback?” speeds up design reviews faster than any Figma annotation.
The Hardest Part (That Nobody Talks About)
Finding users to interview every week.
This is the practical friction that kills most continuous discovery programs before they start. You email your user panel, you get three responses, two reschedule, and suddenly it’s Thursday and you have no one to talk to.
The fixes:
- Put a Calendly link on your main onboarding flow (“Got 20 minutes? Share your first impression.”)
- Ask every customer success conversation to close with a research referral
- Build a lightweight research repository that captures opt-in interview candidates by segment
At Voxy, we built a simple Airtable base to track interview candidates by user type, proficiency level, and recency of last interview. What looked like a scheduling problem was really a funnel problem — and product teams are good at fixing funnels.
The Biggest Misconception About Continuous Discovery
People hear “talk to users weekly” and think it means collecting feature requests weekly.
It doesn’t.
You’re not asking users what they want. You’re watching what they do, asking why they did it, and listening for the emotion underneath. “I just gave up” is worth a thousand survey responses. The opportunity hides in that feeling — the gap between what the user was trying to accomplish and what your product actually helped them do.
Teresa Torres calls this the “underlying job.” You’re not building features; you’re closing gaps.
Is This Right for Every Team?
No. Continuous discovery works best when:
- Your users are accessible (consumer or SMB products, not enterprise-only)
- You have at least a product trio with regular touchpoints
- Leadership tolerates discovery time in sprint planning
It’s harder in heavily regulated industries, in enterprise sales cycles, or in teams where research is siloed in a UX department with a six-week turnaround. But even there, the habit can be approximated — with internal stakeholder interviews, with sales call recordings, with support ticket deep dives.
The point isn’t the method. The point is the cadence.
Start With One Interview
If you want to try continuous discovery without committing to the full framework: book one user interview this week. Not for a specific project. Just a 20-minute conversation about their experience with your product.
Take notes. Share them with your team.
Do it again next week.
After four weeks, something shifts. You start thinking about users differently. You start hearing their words in roadmap debates. You start noticing when a decision is speculation and when it’s grounded.
That shift is the whole point.
This article is drawn from my experience implementing continuous discovery at Voxy, where a weekly interview cadence cut our discovery-to-decision cycle in half. For the full case study, see Continuous Discovery at Voxy.