To persevere with a struggling product, teams need two things: deep conviction that the user problem is important, and a single instance where the product works flawlessly. This "perfect golf shot" provides a glimpse of its potential and the motivation to continue.
To maintain speed, leaders in large companies should focus their personal energy on high-potential projects that the organization won't solve on its own. These are often risky, cross-functional initiatives that require senior intervention to overcome corporate inertia.
Stories succeeded not because it was a new format, but because it solved a core Instagram user problem: the pressure to post only "perfect" photos. It created a "pressure release valve" for casual, ephemeral sharing, making it a natural fit that unlocked latent demand.
Instead of focusing on a slowly declining retention curve, look for the curve to flatten or even tick upwards over 30-90 days. This "J-curve" indicates that a core group of users is forming a stable habit, a stronger signal of PMF than initial user numbers.
Two of Instagram's biggest features were initial disasters. Reels was buried in Stories, and Close Friends was confusing. They were saved by the team's conviction in the core user need, which fueled the persistence required to iterate past the failed first versions.
The biggest problem in buying a TV isn't the final click to pay, but the hours of research. An effective AI agent should handle all the context-gathering (room size, reviews, deals) to present a highly informed choice, super-charging the user's decision rather than replacing it.
Robby Stein argues against making AI the default search experience, as many queries are simple and don't benefit from it. The best approach is a unified system that intelligently surfaces AI previews for complex questions, allowing users to dive deeper without a disruptive mode switch.
Unlike passive consumption apps, where getting many users to try a feature once is key, high-intent products like Google Search measure success by user intensity. The critical question is not "how many people used it?" but "are individual users using it more intensely over time?"
Even for Google, new products start with a small group of trusted testers. The key turning point isn't a metric but a qualitative signal: when early users go from reporting bugs to proactively sharing stories about how the product solved a complex problem for them in an unexpected way.