The lapse isn't the problem. The collapse is.

There’s a strange finding buried in old dieting research. Put someone on a strict diet, convince them they’ve already blown it for the day — a milkshake, say — and then leave them alone with a bowl of snacks. The dieter doesn’t eat less than the person who never dieted. They eat more. The one who was trying hardest to restrain themselves overshoots the hardest, the moment they believe the streak is broken.

Researchers called it counterregulation, and later gave the everyday version a blunter name: the what-the-hell effect. Once a goal framed as all-or-nothing gets violated even slightly, the behavior doesn’t drift back toward the line. It falls off a cliff. “I’ve already broken it, so it doesn’t matter now.” Relapse researchers found the same shape and called it the abstinence-violation effect — the single slip isn’t what predicts a full return to the old habit; the reaction to the slip is.

Sit with what that means for anything you build with a streak, a goal, a budget, or a daily loop in it. You have quietly split your user’s experience into two very different events, and you’ve probably been optimizing the wrong one.

The first event is the lapse: they miss a day. Behaviorally, this is almost nothing. One skipped session out of forty is a rounding error against the habit you’re actually trying to build.

The second event is the collapse: the story the user tells themselves right after the lapse. The streak is gone. The record is ruined. I’m not really a person who does this. That story is the churn event. Not the missed day — the meaning of the missed day.

Most streak designs actively manufacture the collapse. A big number that resets to a hard zero the instant it breaks is an all-or-nothing frame — you built the milkshake right into the mechanic. The user who misses one day doesn’t see “39 out of 40.” They see a shattered zero, and the brain fills in the rest of the sentence.

So the build decision isn’t “make the streak more motivating.” It’s engineer the re-entry. Three moves, roughly in order of leverage:

Reframe progress from a record into a trend. A perfect record is a single point of failure; an accumulation isn’t. “You’ve shown up 39 of the last 45 days” survives a miss. “Day 0” doesn’t. What you count, and whether a single gap can zero it, is a design choice about which story the user gets to tell.

Build in forgiveness before the lapse happens. The streak freeze is now common lore — a bank of grace days that absorb a miss so the count dents instead of dies. (Hold the specific example loosely; the mechanism is the point, not any one product’s version of it.) You’re not being generous. You’re refusing to detonate a habit over one bad Tuesday.

Own the moment right after the miss. That first session back is the highest-stakes screen in the whole loop, and most products spend it on silence or a guilt-shaped notification. It should do the opposite of what the what-the-hell voice is saying: you missed one — the forty you built are still here — pick up where you were.

The uncomfortable part is that the collapse is easier to trigger than to prevent, and a naive “keep ‘em accountable” instinct triggers it on purpose. The strongest-willed user — the one restraining themselves hardest to keep the perfect record — is exactly the one who overshoots into quitting when it breaks. Design for the slip you know is coming, and the lapse stays a lapse.

The science, to look up: the “what-the-hell effect” (Cochran & Tesser, 1996); counterregulation and disinhibition among restrained eaters (Herman & Polivy); the abstinence-violation effect in relapse-prevention work (Marlatt & Gordon). These come largely from eating and addiction labs — the extension to product streaks and goals is by analogy, so treat the size of the effect as suggestive, not settled.