Superhuman's product team scored 22% on Sean Ellis's product-market-fit benchmark the first time they measured it: the share of users who said they'd be "very disappointed" without the product, against Ellis's 40% threshold for having found real fit. What closed that gap wasn't a redesign. According to Rahul Vohra's account in First Round Review, the team segmented respondents by their survey answer, then interviewed the "somewhat disappointed" group specifically to learn what would move them into the "very disappointed" camp. That interview-driven segmentation pushed the score to 33%, then to 58% (First Round Review, 2018-11-13).
The transferable lesson isn't the benchmark number. It's that the interview was built to answer one question the team would act on immediately, not to accumulate impressions. Most founder interviews fail at exactly that step. Teresa Torres, in her Continuous Discovery Habits framework, names the specific anti-patterns that turn interviews into notes instead of decisions: one person on the team owns all the recruiting and interviewing, findings get shared as pages of raw transcript or recordings instead of synthesized takeaways, and the team stops its whole build process to batch-process a pile of interviews after the fact (Product Talk). Her prescription is structural, not motivational: continuous discovery means weekly touchpoints with customers, run by the people actually building the product, each one tied to a specific desired outcome rather than open-ended "tell me about your day" exploration (Continuous Discovery Habits, Teresa Torres / Product Talk, 2021-04-14).
Ask about the problem, not the solution
Rob Fitzpatrick's The Mom Test gives the rule that keeps an interview from turning into a pitch meeting: "You aren't allowed to tell them what their problem is, and in return, they aren't allowed to tell you what to build. They own the problem, you own the solution" (The Mom Test, Rob Fitzpatrick, 2013-09-10). Founders violate this constantly, usually by asking "would you use a feature that does X" and recording the polite yes as validation. The fix is mechanical: ask what the customer already does today, what they tried before, what it cost them in time or money, and stop there. Fitzpatrick's second rule is a useful gut check on whether you're actually doing this: "You should be terrified of at least one of the questions you're asking in every conversation" (same source). If every question in your interview guide is comfortable to ask, you've written a customer satisfaction survey, not a discovery interview.
Interview the switch, not the wish
Intercom's Jobs-to-be-Done work with Bob Moesta reframes who you should be interviewing in the first place. Instead of asking current or prospective users what they want, the switch interview targets people who recently bought or recently canceled, and asks a single causal question: why was today the day they signed up (or quit)? (Intercom Blog / Bob Moesta podcast). That question forces a timeline: what was happening the week before, what triggered the search, what almost stopped them. It produces a specific causal chain you can build a positioning or onboarding decision around, instead of a wish list you can't prioritize. This is the same operating principle Steve Blank stated flatly at the start of the customer development movement: "There are no facts inside the building, so get outside" (Steve Blank, 2009-10-08). The facts aren't in your product analytics or your own head. They're in the specific moment a real customer decided to act.
What to do this week
Pick one pending product or positioning decision you're currently guessing on. Book three interviews, each with a customer who just did the relevant thing (bought, churned, upgraded, or ignored your last email). Ask only about their situation before and the moment they acted, not about your roadmap. Write down the one decision each interview should settle before you take the call, and don't let the conversation end until you have an answer to it, even an uncomfortable one. If none of your three questions makes you nervous to ask, rewrite them before the next round.