How to break a bad habit: why willpower is the wrong tool

This page is about habits you want rid of. The late-night scrolling, the crisps in front of the television, the drink on a Tuesday, getting into the car when the bike is right there. Making a new habit stick is a different question, and it lives in the guide to staying motivated to exercise, along with how long it takes and what a missed session really costs.

A habit is a situation, not a character trait

The most frequently quoted estimate of how much of life runs on habit comes from a diary study published in the Journal of Personality and Social Psychology in 2002. Participants wore a wristwatch that beeped once an hour. When it beeped, they wrote down what they were doing and what they were thinking about at that moment. For each behaviour they also rated how often they usually did it, and how often it happened in the same place.

Behaviours the participants reported performing just about every day and usually in the same location were classed as habits. That came to 35 per cent in the first study and 43 per cent in the second. The researchers summed it up as between a third and a half of everything the participants did.

Two caveats almost always get left out when the figure is quoted. The first: the participants were American college students, which is not a group whose daily life looks much like that of a 45-year-old with a job and children. The second: the 40 per cent figure often quoted alongside this study applies to habitual behaviour only. Even in the middle of a habit, participants' thoughts were about what they were doing around 40 per cent of the time. For the rest of it they were thinking about something else.

That has a practical consequence, and it is the foundation for the rest of this page. The question why am I so weak goes nowhere, because it hunts for the fault inside you. The question that goes somewhere is: where am I standing, what did I just finish doing, and what can I see when it happens?

Why willpower is the wrong tool

The most heavily researched technique for steering your own behaviour is called an implementation intention, and it takes this shape: if situation X occurs, then I will do Y. If I walk through the kitchen after dinner, then I fill the water bottle. A British research group wanted to know why the technique works, and published the answer in the British Journal of Social Psychology in 2008.

The first part was a meta-analysis of 66 tests. It asked whether the plan works by making people more motivated or more confident in their own ability. The answer was no. Forming an if-then plan had negligible effects on both of those.

The second part showed what actually carried the effect: how accessible the specified situation became in memory, together with how strong a link was forged between that situation and the intended response. The plan is not working on your will. It is working on your cue.

That also explains why habits break in the evening, after a bad night, or in the middle of a heavy stretch. Willpower is the resource that disappears when you are depleted. The cue is still standing exactly where it always stood. What to do with stretches where everything is depleted is in the guide to stress management.

What happens when you actually move things around

If the cue is the lever, the next question is how much it can lift. The best summary is a Cochrane review from 2019 on altering the availability or the proximity of products. Twenty-four randomised studies were included. Despite the title, every one of them was about food: not a single one about alcohol or tobacco. Fourteen were run in a laboratory, fourteen were run in the USA, and twenty of the twenty-four raised concerns about risk of bias.

SMD means standardised mean difference, in other words the size of the effect expressed in standard deviations. Around 0.5 is usually described as moderate. The certainty rating is the review's own assessment using GRADE.

The direction is consistent throughout: further away means less, fewer options means less. But look at the third row. Its confidence interval crosses zero, which means that particular effect is not established.

And then the finding that ought to make anyone cautious. The review's own analysis of what explained the differences between studies showed that the proximity effect was larger in the studies that were at high risk of bias. That is the signature of an effect partly manufactured by sloppy methods. The authors still concluded that these changes could contribute to meaningful shifts in behaviour, while noting that the certainty is low or very low and that what is needed is research in real-world settings over sustained periods.

The specific application to sweet things, meaning what the distance to the bowl does to your evening, is in the guide to sugar cravings at night.

The cold shower of 2022, and the one thing left standing

In January 2022, PNAS published the largest review to date of nudging, meaning the practice of designing choices so that the desired option becomes the easy one. It drew on more than 200 studies with over 450 effect sizes and just over 2.1 million participants. The result was an effect size of 0.45, with a confidence interval running from 0.39 to 0.52. The paper concluded that nudges are an effective and widely applicable tool.

In July of the same year, the same journal published a reply from researchers in London, Amsterdam and Australia. Their objection was that the first paper had itself reported moderate publication bias, and that this was the real headline. They reran the analysis using a method that combines several different models of how the distortion might work.

After that correction, no evidence remained for an overall effect. For the category built on informing people, there was evidence against an effect. For the category built on supporting an intention, meaning reminders and planning aids, there was also evidence against. For the category that changes how the options are arranged and which ones exist at all, the verdict was inconclusive.

The conclusion is not that nothing works. It is that you should spend your effort on the one category that survived the scrutiny, and that you should expect less from it than the internet promises.

An example of what survived: the wine glass

In January 2024, PLoS Medicine published a field trial across 21 pubs, bars and restaurants in England. The design was a reversal trial over three four-week periods: normal range, altered range, normal range again.

The change was minimal. The largest glass of wine was taken off the menu, meaning 250 millilitres in 18 of the venues and 175 millilitres in three. Nothing else was done. No signs, no advice, no information given to customers.

During the period with the largest glass removed, 420.8 millilitres less wine was sold per venue per day, with a confidence interval running from 160.2 to 681.4 millilitres. That works out at 7.6 per cent. Sales of beer and cider showed no detectable change, nor did daily revenue, though the study was not powered to settle that particular question.

No customer was informed. No customer decided anything. The largest glass simply was not on the list any more. That is the kind of change that survived the scrutiny, and it is the kind of change you can make at home. Do not resolve to eat fewer crisps. Change what the largest available portion is.

The window that opens when life changes

Because the habit lives in the situation, something interesting happens when the situation is swapped out. The phenomenon is called habit discontinuity: during a move or another change of context, the old cues lose their grip for a while, and the behaviour has to be thought through again.

The best controlled test is a field experiment published in the Journal of Environmental Psychology in 2016. Eight hundred residents of Peterborough in the east of England took part. Half had moved house within the previous six months. Half received a visit with advice on more sustainable everyday habits, half received nothing.

The improvement showed up in one group only: those who had lived in their home for between 1 and 13 weeks and who also received the intervention. Neither the intervention on its own nor the move on its own was enough. The window therefore looks to be about three months long.

Earlier work from 2008 pointed the same way. Among university employees, those who had recently moved and who cared about the environment drove to work less often than equally concerned colleagues who had not moved.

The honest caveat: both studies are about recycling and travel choices, not about exercise or food. Nobody has shown that the window looks the same for a training habit. The mechanism is the same one the rest of this page rests on, though, which is that the answer stops arriving automatically once the cues are replaced.

In practice this makes timing a resource. A new job, a house move, children changing school, the start of a term, the end of the summer holidays: that is when change is cheapest. New Year's Eve is a poor moment for exactly the same reason, because everything else in your life looks identical on the second of January. How to rebuild a week that is already full is in the guide to having no time to exercise.

Field manual: six moves in the right order

1. Map the situation, not the habit. The next three times it happens, write down five things: the time, the place, what you were doing immediately before, who you were with, how you felt. Three notes are almost always enough for the pattern to become visible. You are looking for the cue, not for an explanation of your character.
2. Put one more physical step between you and the habit. A single extra movement goes further than a promise. The phone charges in the hallway. The crisps never come home from the shop. The bike helmet sits on top of your shoes. The aim is not to make the thing impossible, only inconvenient.
3. Shrink the largest available portion. That is the lesson of the wine glass trial. A smaller bowl, a smaller bag, a smaller glass, a shorter time limit that is already set. You still choose freely, but you are choosing from a different range.
4. Give the cue a new answer. An empty slot gets filled by the old thing. Decide in advance what happens instead, phrased as an if-then: if I sit down on the sofa once the children are asleep, then I put music on and tidy up for ten minutes. The replacement has to be as close to hand as the thing it replaces.
5. Put the change inside a context shift if you have one coming. If you do not, one will arrive sooner or later. Start now anyway, and rebuild around the shift when it arrives. Do not wait for New Year.
6. Count in weeks, not in consecutive days. An unbroken chain is a fragile measure, because the only thing it can do is break. Ask instead how this week compared with the same week a month ago.

Four common mistakes

• The ban. Deciding that something must never happen again usually makes it more charged rather than less. Why that backfires specifically with sweet things is in the guide to sugar cravings at night.
• Relying on the evening. Most habits you want to break fire late, which is exactly when willpower is at its lowest. Rebuild the evening in the morning, while you still have something to build with.
• Changing everything at once. Every change costs attention, and attention is shared across all your simultaneous projects. Take one habit at a time.
• Judging yourself on a relapse. A relapse means a situation you had not mapped turned up. That is information about your environment, not a verdict on you.

When it is not a habit

Most of this is ordinary everyday behaviour. Sometimes it is something else, and that something deserves real help rather than more tricks.

The line sits roughly where control runs out. A habit is inconvenient to break. Dependence means you carry on despite clear negative consequences for your health, your finances, your work or your relationships, often alongside strong cravings and withdrawal symptoms when you go without. A compulsion is something you have to repeat in order to damp down anxiety, where the anxiety climbs sharply if you resist.

If it is your energy that has gone instead, and has been gone for a long time, that is a different thing to look at. It is covered in the guide to burnout recovery.

Frequently asked questions

How do you break a bad habit?

By changing the situation that triggers it, rather than by deciding harder. Map first where and when the habit fires, meaning the time, the place, what you were doing immediately before, and how you felt. Then put one extra physical step between yourself and the behaviour, shrink the largest available portion, and decide in advance what you will do instead when the cue arrives. That is the order the research supports: changing what is in front of you held up when the evidence was scrutinised, while informing yourself and reminding yourself did not.

How long does it take to break a bad habit?

There is no single number that applies to everyone, and the number you most often see quoted is weaker than it sounds. How long it takes to install a new habit is covered in the guide to staying motivated to exercise, where both the origin of the 21-day myth and the actual research can be found. In practice, weeks are a better measure than consecutive days. An unbroken chain can only ever break, whereas the question of how this week compared with the same week a month ago can still be answered honestly after a relapse.

Why do I fall back into the habit when I am tired or stressed?

Because willpower is the part of the system that disappears when you are depleted, while the cue is still standing exactly where it always stood. That is also why most habits people want to break fire in the evening. The practical consequence is that the evening has to be rebuilt in the morning, while you still have something to build with. A relapse usually means a situation you had not mapped turned up, which makes it information about your environment rather than a verdict on you.

Is it enough just to move things around at home?

It helps, but expect a modest effect. The Cochrane review from 2019 found that food placed further away was eaten in smaller amounts, with a standardised mean difference of −0.60, and that fewer options reduced what was selected. Certainty was rated low in both cases. The review's own analysis showed that the proximity effect specifically was larger in the studies that were at high risk of bias, which is reason to read the numbers as a direction rather than a promise. The direction is still the best we have, because this was the only kind of change left standing when the nudge research was corrected for publication bias.

Is a bad habit the same thing as an addiction?

No. A habit is inconvenient to break. Dependence means you carry on despite clear negative consequences for your health, your finances, your work or your relationships, often alongside strong cravings and withdrawal symptoms when you go without. A compulsion is something you have to repeat in order to damp down anxiety, where the anxiety climbs sharply if you resist. If you recognise yourself in the description of dependence or compulsion, take it to a doctor rather than hunting for more tricks. In Sweden you can call 1177 for round-the-clock health advice.