Every decision, and why it was made that way
Most of what follows is a choice someone else made differently. Where that is the case, the reason is stated — so you can disagree with it on the merits rather than taking it on trust.
What it does
Six things most habit trackers get wrong
Each of these is a decision with a reason behind it, not a feature list.
Habits that forgive a bad week
A day counts at 60% of its expected load, not 100%. A habit due four days a week only asks for four. Six of eight glasses of water is 0.75 of that habit, not a failure. Streaks you can actually keep.
Read moreSleep, in stages
Duration, deep, REM, light and awake, with bedtime consistency tracked separately — because when you go to bed predicts more than how long you stay there. Ready for Apple Health and Health Connect.
Read moreA journal that stays yours
Morning intentions, evening reflections, and an AI that reflects on what you wrote. The text never leaves your account — not to an ad platform, not into a training set, not into our own dashboards.
Read moreStatistics, not vibes
Pearson and point-biserial correlations with exact t-distribution p-values. A finding appears only if it clears three floors at once: 20 paired days, |r| ≥ 0.25, and a corrected p ≤ 0.00625.
Read moreHonest about nothing
The hypotheses are fixed in advance rather than trawled, so "nothing to report" is a real outcome the screen is designed to show. Every card carries its r and its sample size.
Read moreCalm by construction
No streak-loss guilt, no red badges, no notification you did not ask for. Every colour in the interface was contrast-solved rather than eyeballed, in both light and dark.
Read more
What you actually see
Your day, then the pattern behind it
Two screens carry the product. Today is what you open in the morning. Insights is what you read at the end of a month.
Today
Whole habits finished, so the ring always agrees with the checkmarks.
Insights
Every claim about your life arrives with the evidence for it.
Better sleep, more training
r = 0.41 · n = 118 · p < 0.001On nights over 7h 20m you trained on 71% of the following days, against 34% otherwise.
Morning meditation lifts tomorrow
r = 0.29 · n = 104 · p = 0.003Mood averaged 0.5 higher the day after you sat, controlling for sleep.
Illustrative figures from the demo dataset — the same ones the seeded history produces, and the same filters your own data passes through.
How it differs
Compared with a typical habit tracker
Behaviours rather than brands, so you can check each one yourself.
| Capability | Crescendo | Typical |
|---|---|---|
| Streaks that survive one missed day | Yes | No |
| Partial credit on numeric targets | Yes | No |
| Correlations with p-values and sample sizes | Yes | No |
| Pre-registered hypotheses, not data trawling | Yes | No |
| Sleep stages alongside habits | Yes | Varies |
| Journal text excluded from all analytics | Yes | No |
| Full data export and deletion, self-service | Yes | Varies |
| Contrast-audited in light and dark | Yes | No |
Give it a month and see what it finds
Twenty paired days is the minimum before the statistics will say anything at all. That is roughly three weeks of ordinary use.
Or leave an email and we'll tell you when the mobile apps land.