Behavior-Driven Product Development
Watch what customers do, not what they say, and let them hijack your roadmap
There is often a vast divide between what customers say they want and what they actually do. If you follow their suggestions too literally, you can find yourself without followers. Google discovered this foundational principle when users said they wanted 30 search results per page but actually preferred 10, because the barely perceptible loading speed difference of showing more results caused measurable drops in engagement.
The most powerful product insights come from observing how customers actually behave, especially when they 'cheat' or use your product in unintended ways. Rent the Runway discovered its subscription model when customers kept wearing rented cocktail dresses to work the following Monday. ClassPass discovered its unlimited model when users forged new accounts to get additional passport credits. Dropbox discovered critical usability flaws only when they watched users fail to complete basic tasks in person.
The framework prescribes treating your customers as scouts who push the frontiers of your product and bring back intelligence. You watch, you follow, and you let them lead you to opportunities you would never have found through surveys, focus groups, or your own imagination.
- People are very poor at predicting their own reactions to new things
- Surveys tell you what people think they want; behavioral data tells you what they actually value
- When users 'cheat' your system, they are revealing the product they wish you had built
- Release early, observe behavior, react, and repeat. This cycle is more valuable than extensive pre-launch planning
- The key to observation is keeping an open mind rather than confirming hypotheses you already hold
- Universal customer challenges you discover through observation become scalable product opportunities
- Set up behavioral observation alongside surveysNever rely on stated preferences alone. For every survey or focus group, design a complementary behavioral test. Google's experiment framework deployed different versions of the same page and measured actual usage patterns. Dropbox ran 'poor man's usability tests' by offering Craigslist respondents $40 to complete basic tasks while being observed.
- Watch for the cheaters and workaround artistsMonitor how users bend, break, or circumvent your product's intended use patterns. Rent the Runway customers wearing rented cocktail dresses to work. ClassPass users forging new accounts for extra credits. These 'cheats' are the highest-signal product feedback you can get because users are investing effort to get something your product does not yet offer.
- Build what the behavior tells you, not what the complaint saysWhen you see behavioral patterns that diverge from stated preferences, build for the behavior. Rent the Runway pivoted from single-event rentals to a subscription 'closet in the cloud' because behavior showed customers wanted everyday fashion access, not just special occasion dresses. This became a much bigger business than the original concept.
- Test in the real world, not just digitallyEventbrite invested in RFID technology for large events, but only by observing it in use at actual festivals did they discover that the chip readers were embedded in immovable gates that created bottlenecks. They then built a small clamp-on reader, but field observation revealed it required a wrench to move. Three iterations of physical observation produced the final solution.
- Let users lead you to adjacent opportunitiesJulia Hartz of Eventbrite noticed her platform being organically adopted by speed-dating organizers, LEGO enthusiasts, and goat yoga purveyors. Rather than constraining these uses, she studied each new category to understand unique needs and built features to support them. The most visionary founders let customers redefine the product's boundaries.
Jenn Hyman launched Rent the Runway for single-event dress rentals. She noticed customers keeping rented cocktail dresses past their return date and wearing them to work on Monday with a blazer. Rather than enforcing stricter return policies, Jenn recognized that customers wanted confidence in how they looked every day, not just at galas. She pivoted from a la carte rentals to a subscription service allowing multiple pieces to be rotated continuously.
In Google's early days, Marissa Mayer surveyed users to determine the ideal number of search results per page. Users overwhelmingly said 30. But when Google ran A/B tests showing different numbers of results, engagement dropped dramatically at 20 and was worst at 30. The explanation was loading speed: more results meant slightly slower pages, and users cared far more about speed than they consciously realized. This experiment became foundational to Google's culture of measuring behavior over stated preferences. Mark Zuckerberg observed the same pattern at Facebook: users complained about every expansion (to new campuses, photo tagging) but their actual usage always increased.