Storytime
Day 41. You are ill, or away, or it is half past eleven and you have only just sat down. You do not do the thing. In the morning the app says 0.
A streak is a number that only goes down. That is the whole problem, and everything else in this article follows from it. It goes up by one a day, slowly, for as long as you are perfect, and the moment you are not it goes to zero in a single step. Forty days of work and one bad evening are worth the same to it, which is nothing. No other measure you willingly look at behaves like this.
Streaks are not useless. They are extremely good at one job and actively harmful at another, and almost every habit app ships them without distinguishing between the two.
What a streak measures, and what it does not
A streak measures unbroken consecutiveness. It answers one question: have you done this every single scheduled day since the last time you did not.
That is a real question, and while the answer is yes it is a good motivator. The reason is well documented. Kivetz, Urminsky and Zheng revived the goal-gradient hypothesis in the Journal of Marketing Research in 2006, using, among other things, a cafe loyalty card: customers bought coffee faster the closer they got to the free one. Progress towards a visible target accelerates effort. A streak is a target you can see, so it pulls.
The same paper contains the part nobody quotes. After customers collected the reward, their rate of purchase dropped, and then climbed again as the next reward came into view. Effort tracked the number rather than the coffee. That is fine for a loyalty card. It is a worrying property for something that is meant to be building a behaviour you will still have in five years.
And a streak answers only that one question. It cannot tell you whether the habit is getting easier, whether it survives travel, whether you come back after a gap, or whether the last 30 days were better than the 30 before. Those are the things that describe a habit. Consecutiveness is a constraint that has been mistaken for a measure.
Worse, what the streak punishes is the one thing the research says is harmless. In Lally, van Jaarsveld, Potts and Wardle's study of 96 people repeating a daily behaviour for 12 weeks (European Journal of Social Psychology, 2010), automaticity climbed along a curve that plateaued after a median of 66 days, with a range of 18 to 254. Missing a single opportunity did not meaningfully affect that curve. The habit carried on from where it was. The streak is the only thing in the room that went to zero, and it is the thing you were looking at.
The two failure modes
The reset spiral. One miss, then the number is gone, then the story starts. Larimer, Palmer and Marlatt described the mechanism in the relapse literature as the abstinence violation effect (Alcohol Research and Health, 1999): after a first lapse, what predicts a full collapse is not the lapse but the explanation, and the dangerous explanation is the one that reads the lapse as a stable fact about you. Cochran and Tesser gave the everyday version its name in 1996, in Martin and Tesser's Striving and Feeling: the what-the-hell effect.
A streak is an explanation generator. It does not say "you missed Tuesday". It says your count is zero, which is the same thing it would say if you had never started, and that is a claim about you rather than about Tuesday. Add Kahneman and Tversky's finding that losses weigh more than equivalent gains (Econometrica, 1979) and the asymmetry is complete: 40 days accrued a point at a time, lost in one.
The hollow streak. The other failure mode is quieter. The number becomes the goal, so it gets protected: a tick for a run that was a walk to the corner, a tick at 23:58 for a page you did not read, a tick for a session you shortened until it no longer resembles the habit. The record stays clean and stops being true, which makes it useless for the one thing a record is for.
Be careful with this one. It is widely reported and, as far as I can find, has never been measured: there is no study of tick-inflation in habit apps. Believe it because you have done it, not because somebody has a figure.
Who streaks actually work for
The honest answer is: the people who do not miss.
If your habit is small, daily, tied to a cue that fires without you and immune to travel, a streak is a clean and pleasant piece of feedback, and you should keep it. It will run for a year, it will feel good every day, and none of the failure modes above will ever apply to you.
That is the problem, though, put plainly. Streak mechanics work best for the people whose habit is already established and worst for the people still trying to establish one, which is the opposite of how you would design it if you started from the user. The design serves the people who need it least, and punishes hardest at precisely the moment, the first miss in the first month, when the habit is most fragile. That moment has its own article: why you give up on new habits in week two.
There is a second group it works for: short, bounded challenges. Thirty days of something, deliberately finite, where the streak is the point and the ending is scheduled. Nobody is harmed by that. It is the open-ended, indefinite streak on a habit you intend to keep for life that eventually meets a bad week and takes the habit with it.
What to track instead
Three measures, all of which survive a missed day.
Rate over a window. How many of the scheduled days in the last 30 did you do. This is the number that most closely matches what you actually want to know, it degrades gently rather than collapsing, and it makes the arithmetic obvious: 27 out of 30 is excellent, and it is also a month with three misses in it, which a streak would have rendered as zero, zero and zero.
Time to recover. How long the gaps are. One day off and straight back is a different habit from one day off and a fortnight of nothing, and a streak counter records both as the same event.
This is the measure with the best evidence behind it, and it is worth being specific about why. In 2021 Milkman and colleagues published a megastudy in Nature: 61,293 members of an American fitness chain, 54 different four-week programmes designed by 30 scientists across 15 universities, all measured the same way at the same time. Of the 54, 45% significantly increased weekly gym visits, by 9% to 27%. The best of all 54 was a small bonus, 125 points worth about nine cents, paid for returning to the gym after a missed workout. It produced 0.40 extra visits a week, a 27% increase over the control.
The most valuable moment in that entire experiment was the day after a miss. A streak assigns that moment a value of zero. That is the argument of this whole article in one line.
Reasons. Why the gaps happened. Three missed Fridays with "away" against them is a scheduling fact and tells you to change the schedule. Three missed Fridays with nothing against them is a mystery, and mysteries do not get fixed. This is the least glamorous of the three measures and probably the most useful, because it is the only one that tells you what to do next.
The honest caveat: nobody has run a trial comparing a tracker that shows a streak against a tracker that shows a recovery rate. The case above is assembled from habit-formation research, relapse research and one very large field experiment on gym attendance, not from a head-to-head test of app designs. Anybody who tells you otherwise is selling something, quite possibly a habit app.
If you want a record that treats a missed day as a missed day, Seeded does that. It keeps streaks as a record rather than a scoreboard, and a skip can carry a reason, Rest day, Unwell, Away or Busy, so the gap in the record says what it was. It is free for three habits on iPhone.
If your app has a streak, how to hold it
You do not have to leave. You have to demote it.
Treat the number as a record of a run that happened, not as a score you are defending. A streak of 40 that ended is a true statement about 40 days, and it stays true afterwards. The counter says zero because that is the only thing a counter can say; it is not a summary of you, and it is not a summary of the habit either.
Then, practically: look at the month rather than the day. Count the ticks in the grid instead of reading the headline figure. Write down why the gaps happened. And on the day after a miss, do the habit smaller than usual, which is the entire subject of what to do the day after you break a streak.
One more thing worth knowing: when a habit is going well it stops being interesting, and the streak is often the only thing still supplying a feeling. That is a bad reason to keep a habit and a good sign it has actually worked, which is the argument in the boring routine is the point. Small wins are a real motivator, and Amabile and Kramer's diary study of 238 people across 26 project teams found that progress in meaningful work was the strongest driver of a good day at work (Harvard Business Review, May 2011, and their book The Progress Principle). That was offices rather than habits, so take it as a framing rather than a result, but the framing is right: the win you want is the return, not the run.
What "don't break the chain" was actually about
The method is usually attributed to Jerry Seinfeld: get a wall calendar, put a big red X on every day you write, and after a while your only job is not to break the chain. The story comes from a 2007 Lifehacker piece by Brad Isaac, who says he was given the advice by Seinfeld in a comedy club.
Seinfeld has publicly disowned it. Asked in a Reddit AMA on 6 January 2014 whether he still put a big X on his calendar every day, he answered: "This is hilarious to me, that somehow I am getting credit for making an X on a calendar with the Seinfeld productivity program. It's the dumbest non-idea that was not mine, but somehow I'm getting credit for it." So the most-cited authority for streak-keeping is a man who says it was never his idea.
What is worth keeping from the original is not the chain. It is the advice underneath it, which was about writing every day: the output was not the point, the showing up was. Somewhere between a comedian's wall calendar and a push notification, the record of showing up turned into a score that could be lost, and a tool for reducing pressure became a machine for generating it.
A habit record is meant to answer one question honestly: what did I actually do. Anything that makes you want to answer it dishonestly, or makes you avoid opening it after a bad week, has stopped being a record. If you are still deciding what a tracker should do for you, that question is also the heart of habit tracker vs to-do list, and the timescales that make all of this easier to sit through are in how long habits actually take.
Seeded is a habit tracker for iPhone that does not shout. Streaks are kept as a record, not a scoreboard, and Pro adds an Insights view that shows how quickly you come back after a gap, which is the measure this article argues for. Three habits, the journal and time capsules are free. Get it on the App Store.
Bitesized
- A streak is a number that only goes down. It rises one a day and falls to zero in a single step, so 40 days of work and one bad evening are worth the same to it.
- Progress towards a visible target genuinely accelerates effort (Kivetz, Urminsky and Zheng, 2006), which is why streaks work while they are intact. The same study found effort dropped once the reward was collected.
- What a streak punishes is what the research says is harmless: a single missed opportunity did not meaningfully affect the automaticity curve in Lally and colleagues, 2010.
- Two failure modes. The reset spiral, where the miss becomes a verdict (the abstinence violation effect, Larimer, Palmer and Marlatt, 1999), and the hollow streak, where the number gets protected and the record stops being true.
- Streaks work best for people who never miss, which is the opposite of the people who need help.
- Track rate over a window, time to recover after a gap, and the reasons behind the gaps. In a megastudy of 61,293 gym members (Milkman and colleagues, Nature, 2021), the best of 54 interventions rewarded returning after a missed workout.
- "Don't break the chain" is attributed to Jerry Seinfeld, who called it "the dumbest non-idea that was not mine" in a 2014 Reddit AMA.