How to Set Up a Continuous Discovery Habit as a PM
Continuous discovery is the practice of having regular touchpoints with customers — weekly interviews, prototype tests, or usage observation sessions — that feed directly into the product decisions you make. It replaces the project-based research cycle (plan a study, recruit participants, wait six weeks, synthesize) with an always-on signal stream that informs roadmap decisions in real time.
This guide covers Teresa Torres's continuous discovery framework adapted for PMs at growth-stage companies: how to recruit an ongoing pool of interviewees, structure weekly sessions, and connect what you learn to actual product decisions.
Step-by-step guide
Define your target customer segment clearly
Continuous discovery only works if you are talking to the right people consistently. Write a one-paragraph segment definition: role, company size, product usage pattern, and the problem space you are exploring. Example: "B2B SaaS PMs at companies with 50-500 employees who manage a roadmap of 10+ concurrent projects and currently use Linear for ticketing." This definition becomes your recruiting filter.
Build an always-open recruiting pipeline
Add a Typeform or Calendly link to your product's in-app experience — in the settings page, in onboarding emails, or via an in-app banner (tools like Intercom or Pendo support this natively). Offer a $30-50 gift card or account credit per session. Aim for 5-10 candidates in your pipeline at all times. Block 60 minutes every Tuesday afternoon as your permanent interview slot so you can fill it without scheduling overhead each week.
Run 30-minute structured interviews, not open conversations
Start every interview with the same two questions: "Walk me through the last time you [experienced the problem you are exploring]" and "What did you do instead?" These questions surface actual behavior, not opinions. The second 15 minutes explore the specific context: tools used, people involved, decision made. End with "What is the most frustrating part of this process?" Do not show mockups or features in discovery interviews — save that for assumption tests.
Capture insights in a structured Notion or Dovetail repository
After each interview, immediately write a 5-bullet summary: customer segment, problem experienced, current workaround, key quote, and one insight. Tag each insight with the opportunity it relates to (e.g., "Onboarding complexity," "Cross-team visibility"). Over time this repository becomes searchable evidence that backs up roadmap decisions with named customer quotes rather than vague claims like "users have told us..."
Build an Opportunity Solution Tree and update it weekly
Draw an Opportunity Solution Tree (OST): your desired product outcome at the top, opportunities (customer problems) as the second level, solutions as the third level, and experiments as the fourth level. After every interview, ask: does this evidence change where I place opportunities? Does it create a new branch? The OST prevents jumping to solutions before validating that the opportunity is real and significant enough to invest in.
Run one assumption test per week alongside interviews
For your highest-priority opportunity, identify the riskiest assumption: "Users will enter this data manually if we provide the form." Test this with a Figma prototype, a Wizard of Oz test, or a concierge MVP. One test per week compounds: 50 tests per year vs. 3-4 per quarter using traditional research cycles. Tests that fail are more valuable than tests that succeed — they prevent shipping the wrong thing.
Common mistakes
Conducting interviews with pre-existing solutions in mind
Discovery interviews should explore problems, not validate solutions. If you find yourself asking "Would you use a feature that does X?" you have already left discovery. That question belongs in a prototype test. In discovery, ask about past behavior and current workarounds.
Talking to the same three customers every week
Familiarity bias is real. If your weekly interview pool is always the same power users, you will optimize for them and miss segments that matter. Rotate deliberately: one churned user, one new user, one heavy user per month. Your recruiting pipeline should prevent accidental repetition.
Not connecting insights to decisions
Discovery that does not change decisions is theater. After each interview, ask: "Would I prioritize this opportunity differently based on what I learned today?" If the answer is never yes, your discovery is not informing your roadmap.
Making discovery a solo practice
The goal is to create a "shared brain" on the team, not just in your head. Bring your EM or a designer to every third interview. Share interview summaries in Slack every week with a "key insight" headline. Continuous discovery changes team culture, not just PM knowledge.
Tips
Use Calendly with a qualifying question ("How do you currently manage your product roadmap?") to filter out non-ICP candidates before they book
Record all interviews with consent and store them in Dovetail or Notion video embeds so team members who missed the session can watch
Review your Opportunity Solution Tree in your weekly PM sync — it surfaces what you have and have not validated and keeps the team focused on the riskiest unknowns
Set a "discovery metric": number of unique customers interviewed per month. A target of 4-6 is achievable and provides enough signal to make confident decisions.
How Vantage helps
Vantage's context ingestion layer connects your discovery artifacts — interview transcripts, Loom recordings, research docs — directly to your PRD. When you generate a PRD, Vantage grounds it in the customer evidence you have collected rather than starting from your assumptions. Discovery feeds generation; generation clarifies what to discover next.