The app that was competing with kitchens

Patients loved the app. Then, three weeks after bariatric surgery, they stopped opening it — at precisely the point where the real behaviour change was supposed to begin.

MyHabeats supports people rebuilding how they eat after surgery. The early signs were everything a product team hopes for: strong onboarding, warm feedback, good early engagement. Then the cliff, reliably, at week three. Inside the team, the diagnosis wrote itself. Engagement drops, so the product must be missing something. The roadmap filled with candidate features, each one a guess about what would bring people back to the screen.

The work, though, was betting on something more specific than engagement. For the model to hold, a patient standing in their own kitchen — tired, back at work, family around them, old routines humming along — had to change what they ate in that moment. That was the behaviour the entire product depended on. Everything else was scaffolding.

Put that way, the blindspot became visible. The team had been asking what would make people open the app. Nobody had asked where eating decisions were actually being made. The answer was kitchens, family tables, commutes, the fridge at 10pm. Emotional triggers, social pressure, decades of habit — all of it operating in rooms the app was never in. Week three was simply the point where hospital structure faded and ordinary life resumed. The app was competing with a kitchen, and the kitchen was winning.

Seen through that lens, the weakness in the plan was structural rather than a matter of missing features. The model assumed screen time at exactly the moments people were furthest from screens. More features would have meant more of the same bet, made more expensively.

The decision that changed was the roadmap itself. Feature development paused. Across three behaviour sprints, the product strategy shifted from adding things to the app towards supporting the off-screen moments that actually shaped eating — designing for the kitchen, the trigger, the lapse and the recovery, with the app as support rather than stage.

The team reported improved 90-day retention, and the deeper shift was in how product decisions were framed. Retention became the signal to watch; the kitchen became the site of design.

“Genuinely blew us away. Full of ‘why didn’t we think of that?’ moments.”

The behaviour that breaks the work is often the behaviour nobody judged. This one got judged in time.