Continuous discovery habits: how to make customer interviews a weekly habit

Why most product teams only talk to customers when it’s already too late

You’ve been there: six months into building a feature, you finally show it to users, and their reaction tells you everything you needed to know back in January. The problem isn’t that you didn’t do research—it’s that you did it once, at the start, and then went dark. Continuous discovery habits flip this pattern entirely. Instead of big, infrequent research projects, you’re talking to customers every single week, making learning a sustainable rhythm rather than an emergency response.

This approach, developed by Teresa Torres and detailed in her book Continuous Discovery Habits, has become the gold standard for product teams at companies like Spotify, CarMax, and Hope Health. The core insight is simple but radical: weekly customer touchpoints compound into product intuition that no amount of analytics dashboards can replace.

Here’s how to actually make it happen—including what to do when you genuinely can’t get customers on the phone.

Why weekly interviews work (when monthly ones don’t)

The magic number isn’t arbitrary. Weekly customer conversations work because they create three reinforcing effects:

• Pattern recognition happens faster. When you talk to four customers in a month, you hear the same pain point three times and start believing it’s universal. When you talk to sixteen customers, you realize that pain point affects a specific segment—and you know which one.
• You stay connected to current reality. Customer problems evolve. Market conditions shift. A monthly cadence means you’re always working with slightly stale information.
• The team develops shared context. When product trios (PM, designer, engineer) interview together weekly, they stop debating what customers want and start debating how to solve real problems.

Teresa Torres recommends at least one customer interview per week, with the ideal being the entire product trio participating. At Spotify, some discovery teams run two to three interviews weekly during active exploration phases. The key isn’t hitting a specific number—it’s building the habit so that “talking to customers” becomes as routine as checking your email.

How to recruit participants without losing your mind

The most common reason continuous discovery fails isn’t lack of interest—it’s recruitment friction. If scheduling each interview takes three hours of back-and-forth emails, the habit dies within weeks. You need a system that runs on autopilot.

Build an automated recruitment pipeline

The goal is a steady stream of customers who’ve already agreed to talk, scheduled themselves, and shown up ready. Here’s the infrastructure:

1. Create a research opt-in at key product moments. Add a simple question after onboarding, after a feature interaction, or in your product’s help section: “Want to help shape the future of [product]? Join our customer feedback panel.” Aim for a form that takes under 30 seconds.
2. Use a scheduling tool with automation. Calendly, SavvyCal, or Cal.com can send reminders, handle timezone conversion, and reschedule no-shows automatically. The best setup includes a reminder 24 hours before and one hour before.
3. Segment your panel from the start. Capture 2-3 qualifying data points during opt-in (company size, role, how long they’ve used your product). This lets you target specific segments when you need them.
4. Offer appropriate incentives. B2C products often need gift cards ($25-50 for 30 minutes is standard). B2B customers frequently want early access to features or a direct line to product leadership. Test what works for your audience.

Intercept-based recruitment for faster results

If you need participants this week, don’t wait for your panel to fill. Intercept recruiting targets customers at the moment they’re doing something relevant:

• In-app prompts: “You just used [feature]. Got 20 minutes this week to tell us about your experience?” Tools like Ethnio or UserLeap specialize in this.
• Support ticket follow-ups: After resolving a ticket, ask if they’d be willing to discuss their broader workflow. These customers are already engaged and often have strong opinions.
• Post-transaction asks: Just completed a purchase? Just cancelled? These are high-signal moments where people are willing to explain their reasoning.

At Intercom, product teams recruit directly from customers who’ve recently used specific features, ensuring conversations are grounded in actual behavior rather than hypothetical preferences.

The interview structure that surfaces real opportunities

Continuous discovery interviews aren’t usability tests or sales calls. You’re not asking “Would you use this feature?” or trying to convince anyone of anything. The goal is understanding the customer’s world—their context, their struggles, and the outcomes they’re trying to achieve.

The story-based interview framework

Torres advocates for story-based interviews that focus on specific past behaviors rather than general opinions. Here’s a structure that works:

1. Start with a specific instance (5 minutes): “Tell me about the last time you [relevant action].” Not “How do you usually…” but “Walk me through what happened on Tuesday.” Specific stories contain details that generalizations hide.
2. Explore the context (10 minutes): What were you trying to accomplish? What had you already tried? Who else was involved? What happened right before? The context reveals constraints and motivations that the customer might not think to mention.
3. Understand the emotional journey (5 minutes): What was frustrating? What felt like a win? Where did you almost give up? Emotion signals importance—if something didn’t make them feel anything, it probably doesn’t matter much.
4. Zoom out to patterns (5 minutes): How often does this happen? Has it changed over time? What would make this easier next time? Now you can generalize—but only after you’ve grounded the conversation in specifics.

Questions that unlock real insight

Keep these in your back pocket:

• “What happened right before that?” (surfaces triggers and context)
• “You mentioned X—can you show me what you mean?” (moves from abstract to concrete)
• “What did you end up doing instead?” (reveals workarounds and alternatives)
• “If that worked perfectly, what would be different?” (clarifies desired outcomes)
• “Who else cares about this?” (identifies stakeholders and buying dynamics)

Avoid asking customers to predict their future behavior or design solutions. “Would you use a feature that…” questions generate unreliable data because people are bad at forecasting their own actions.

From raw notes to usable insights: synthesis that scales

Twenty interviews produce thousands of words of notes. Without a synthesis system, those insights stay trapped in Google Docs that nobody reads. The goal is extracting patterns quickly and connecting them to decisions.

Capture during the interview, not after

Waiting until after the call to take notes means you’re working from memory, which is unreliable. Instead:

• One person leads, one person captures. If your trio is interviewing together, rotate roles. The note-taker focuses on capturing direct quotes and specific behaviors.
• Record with permission. Tools like Grain, Dovetail, or even Zoom’s built-in recording let you clip specific moments later. This is especially useful for sharing compelling customer quotes with skeptical stakeholders.
• Use a consistent template. Include: customer segment, key quotes, observed behaviors, pain points mentioned, and your initial hunches about opportunities. Consistency makes patterns visible across interviews.

The weekly synthesis ritual

After each interview (or at week’s end if you’re batching), spend 15 minutes extracting:

1. Experience map moments: What steps did the customer describe in their journey? Where did they struggle?
2. Opportunities: Unmet needs, pain points, or desired outcomes the customer expressed. Phrase these as customer needs, not solutions: “Needs to understand if their campaign is working” rather than “Wants a dashboard.”
3. Surprises: What challenged your assumptions? What did you expect to hear but didn’t? These are often the most valuable insights.

Connecting interviews to the opportunity solution tree

Continuous discovery habits work best when paired with the Opportunity Solution Tree (OST) [INTERNAL_LINK: opportunity solution tree]—a visual framework that connects your desired outcome to the opportunities you’ve discovered and the solutions you’re considering.

Here’s how weekly interviews feed the tree:

1. Start with your outcome. What metric is your team trying to move? Conversion rate? Retention? Revenue per user? This sits at the top of your tree.
2. Map opportunities beneath the outcome. Each interview surfaces opportunities—customer needs, pain points, or desires that, if addressed, could move your metric. These become branches on your tree.
3. Group and prioritize opportunities. After four to six interviews, patterns emerge. Some opportunities appear repeatedly; others were one-offs. Cluster related opportunities and assess which ones appear most frequently and most urgently.
4. Generate solutions for high-priority opportunities. Only after you’ve validated that an opportunity matters to multiple customers should you start brainstorming solutions. This prevents building features for problems that don’t exist.
5. Design assumption tests. Before building anything, identify the riskiest assumption in your solution and design a small experiment to test it. This might be a prototype test, a painted door, or a concierge MVP.

The tree becomes a living artifact that grows each week. As Lenny Rachitsky often points out, the best product teams treat discovery as ongoing, not as a phase that ends when engineering starts.

What to do when you genuinely can’t talk to customers

Sometimes weekly customer interviews aren’t possible. You’re building for a brand-new market. Your customers are enterprise executives with no time. You’re in a regulated industry with strict communication rules. Here’s how to maintain continuous discovery habits when direct access is limited:

Talk to proxies

If you can’t reach end users, find people who interact with them daily:

• Sales and customer success teams hear objections and pain points constantly. Structure your conversations the same way you would with customers—focus on specific recent examples, not generalizations.
• Support teams know exactly where customers struggle. Pull recent ticket data and interview support reps about the patterns they see.
• Channel partners or resellers often understand customer needs better than internal teams because their income depends on it.

Observe instead of interview

When conversation isn’t possible, observation still works:

• Session recordings (FullStory, Hotjar, LogRocket) show exactly how customers behave, including hesitations and errors they’d never think to report.
• Community forums and review sites contain unfiltered customer language. Mining G2 reviews, Reddit threads, or your own community Slack can reveal opportunities you’d never hear in polite interview conversations.
• Industry conferences and meetups let you observe customers in their natural habitat, discussing problems with peers who face similar challenges.

Run research sprints when access opens

If you can’t maintain weekly cadence year-round, batch intensive research during windows when access is possible. A two-week sprint with daily interviews can build understanding that sustains you through months of limited access.

Making the habit stick

The teams that succeed with continuous discovery habits treat them as non-negotiable rituals, not nice-to-haves that get bumped when deadlines loom. Put interviews on the calendar as recurring events. Protect that time the way you’d protect a leadership review.

Start this week: identify three customers who’ve recently used your product, send a simple interview request, and schedule 30 minutes with one of them. You don’t need a perfect system, a filled research panel, or executive buy-in. You need one conversation—and then another one next week.

The compound interest of weekly customer learning is the closest thing product management has to a cheat code. The only way to access it is to start.

Frequently asked questions