When we started Pixel Learn, we did what every early-stage kids app does: we looked at YouTube Kids, saw the five-minute videos, and copied them. Five minutes is already a good idea, we said. Five minutes is better than twenty-two minutes. Five minutes is a kindness.
Three months later we pulled the data from our first cohort of thirty families. The cohort had watched, on average, 14 minutes of Pixel Learn content per session. The cohort was, on average, four years old. At the end of each session roughly half the kids cried when the screen closed. Two parents asked us to make the videos shorter. One asked us to remove the app.
We had a simple theory: the content was good, kids loved it, so the problem had to be how we ended sessions. We added a warm goodbye screen with the hero waving. We added a "three more videos and then bedtime" countdown. We added parental screentime controls. Crying went up.
The thing we kept missing
The thing we kept missing was this: a five-minute video is not a short video to a four-year-old. It is a world. When the world ends, of course they cry. We were not ending sessions; we were ending universes. No amount of friendly UX chrome can make an ending not feel like an ending.
We weren't ending sessions. We were ending universes. No amount of friendly UX could fix that.
So we did the drastic thing. We cut every video down to two minutes — really two minutes; we cut mid-sentence if we had to. We cut every game to ninety seconds. We put a tiny "what you learned" card at the end of every segment. No countdown, no goodbye screen. Just: here is what you did, the hero winks, the app stops showing you things.
What changed
The numbers above are from our second cohort, eighty families across four countries, six weeks of data. Tantrums at session end dropped by almost two-thirds. Return visits climbed. And — this was the one we didn't expect — average session length shortened. Kids would do one or two two-minute segments and walk away on their own.
It turns out that when you end a universe cleanly, a four-year-old is perfectly happy to start a different universe. Sometimes that universe is blocks on the floor, or a sibling, or a snack. That is, in our opinion, the correct outcome for a kids' app.
Three things we learned the hard way
- The ending is the product. How a lesson finishes matters more than how it starts. We redesigned all our endings before we redesigned any intros.
- Countdowns make it worse. A visible "3 videos left" timer feels like scarcity to a grown-up and like a threat to a child. We deleted it.
- Two minutes is not a constraint. It is a shape. Once we accepted the shape, our writers wrote better scripts, our editors cut tighter, and our games became snappier. The constraint made the work better.
What to try tonight
If screentime ends in a tantrum at your house, try not warning your child. Just let a natural "end" happen — video ends, game ends, done. The warning is what makes the ending feel like a theft. The ending itself is fine.
What we're trying next
In the next release we're experimenting with even shorter segments — 90-second "micro-quests" for kids who are just learning to hold attention. We're also testing a "one more?" button that appears only after a natural stop, so the choice to continue is always made by a calm child, never interrupted out of one.
If you're a parent who wants to join the beta, tell us here. Your kid can be as little as two and as chatty as they want. We promise a short survey, a long thank-you, and zero emails about anything else.