Wondays · redesigned from the habit-formation literature · 2026-08-18

We are all measuring the wrong thing

Atoms, Not Boring and Wondays all count the same variable. The research on habit formation does not measure that variable at all.

What the evidence actually establishes

Four findings

  1. Automaticity grows on an asymptotic curve and then plateaus. Early repetitions produce large gains; later ones produce progressively less, until the curve flattens. Habit formation has a shape, and the shape is known.
  2. Median 66 days to reach 95% of the plateau — but the range was 18 to 254 days. The median is nearly useless as an individual target.
  3. Missing one opportunity did not materially affect the process. Our mercy doctrine stops being a brand value and becomes a finding.
  4. Habit is measured as automaticity, not as frequency. The field's instruments ask whether you do it "without thinking" and "without having to consciously remember" — not how many times you did it.

Lally et al. 2010, European Journal of Social Psychology · the Self-Report Habit Index and its automaticity subscale, Gardner et al. · whether one item can carry it, Gardner 2024 (paywalled — conclusion unverified).

The category error

Behaviour is not habit

Every product in this category counts behaviour: did you do it, how many times, how many in a row. The literature measures automaticity: did you have to decide.

Those two come apart, and the gap is the whole product. A person can rack up two hundred days by grinding out a daily decision and still have no habit at all — the moment they are tired or travelling, it collapses, because it was never automatic. Another person can be genuinely automatic by day forty while their raw count looks unimpressive. The counter says nothing about whether the thing is becoming part of them.

Days won measures what you did.
Nobody measures whether it got easier.
Which makes the 60-day challenge a mistake

The median is not a person

Not Boring's core mechanic is a 60-repetition challenge, drawn from the same body of work. But that work reports a range of 18 to 254 days. A fixed target therefore tells almost everyone something false: it declares victory early for the fast, and manufactures failure for the slow — the very people who already believe they are the type who quits.

You cannot fix this with a better number. There is no number. The only honest target is the individual's own curve, which means the product has to measure the curve.

The redesign

A second atom

Atom one — the record

Did you do it?

One boolean, every day, ten seconds. Unchanged. This is the ledger, the pearls, the year-book — everything we have already built.

Atom two — the becoming

Did you have to decide?

One boolean, asked occasionally rather than daily, phrased in the field's own language. From it: an automaticity curve, that person's own, with that person's own plateau.

It is one extra tap, a fraction of the time, and it is the difference between a tracker and an instrument of becoming. It also finally makes the mission literal: the record shows what you did; the curve shows what you are turning into.

Why occasionally, not daily

Asking every day would double the friction and, worse, force a conscious reflection on a behaviour whose whole trajectory is towards being unconscious. The measurement would interfere with the thing measured.

The consequence nobody else can ship

The app tells you when to stop

When the curve flattens, the honest thing to say is not "level up" or "challenge complete". It is that the behaviour no longer requires the instrument. So the product says it, once, plainly — this has been automatic for six weeks; you do not need to mark it any more — and offers to retire the habit into the record.

Every competitor monetises retention and therefore structurally cannot ship this. Their business needs you to keep tapping. Ours does not, because the value we sell is the record, which only grows, and because a person who graduated one habit is exactly the person who brings the next one.

The only tracker whose success condition is that you stop using it.
And the cue, finally

Own the intention, not the reminder

Stable context is the mechanism of automaticity, and we deleted cues entirely. The fix costs us nothing: the habit is named as the intention"after coffee, move my body" — so the context is captured at setup and the cue lives in their kitchen rather than our push server. One field, phrased differently, and Clear's first law is back without a single notification.

What this changes

Keeps, adds, dies

  1. Keeps: the boolean, the pad, the summary sheets, the string, no streaks, no notifications, no opinion, the year-book.
  2. Adds: the automaticity question, the personal curve, the graduation moment, and the intention-shaped habit field.
  3. Dies: "days won" as the hero number. It stays on the sheet as the record, but the curve becomes the thing you open the app to see — because it is the only number that answers the question people actually have, which is is this working on me yet.
  4. Also dies: any fixed duration, any level, any mastery badge. The plateau is discovered, never assigned.