Consumer sleep apps love a single number: 82, 78, "your sleep score." As someone building in this space, I want to make the case that the score is often the least honest thing on the screen.
Why one number is seductive and wrong
A score compresses a messy, multi-dimensional night - duration, continuity, stage distribution, disturbances - into a scalar you can feel good or bad about. That is great for engagement and terrible for truth. Two very different nights can produce the same 78, and a small model change can move the number without anything about your sleep changing.
What the number hides
- Continuity vs. duration: eight fragmented hours and seven solid ones can score similarly, yet feel nothing alike.
- Confidence: the app rarely tells you how sure it is. A stage estimate from wrist motion is a guess with wide error bars, presented as a fact.
- Events: the actually-actionable stuff - a snoring cluster at 2 a.m., a long awakening - gets averaged away into the score.
A more honest UI
Show the timeline, not the trophy. "Here is when you were restless, here is a disturbance at 03:14" beats "your score is 78" because it is falsifiable and useful. Users can act on an event; nobody can act on a decimal.
The clinical reason the audible events matter - especially for apnea screening - is summarized in this overview of snoring vs. sleep apnea.
How does your product handle the tension between a friendly single metric and honest uncertainty?
Top comments (0)