New Year Resolutions: What the Research Says About Making Them Stick

Almost everything you have read about New Year resolutions rests on a statistic nobody can source — the one where 80% of them collapse by mid-February. This page used to repeat it. The real research is both more encouraging and more useful, and it changes what you should actually do in week three.

Do New Year resolutions actually fail?

Far less than the folklore says. The most-cited study on this followed 159 people who made a resolution and 123 comparable people who wanted the same change but did not — matched on demographics, problem histories and goals — and checked in by telephone over six months.

At six months, 46% of the resolvers were continuously successful, against 4% of the non-resolvers (Norcross et al., Journal of Clinical Psychology, 2002).

Read the second number again. People who wanted to lose weight, start exercising or stop smoking, but did not frame it as a resolution, succeeded at roughly a tenth of the rate. Whatever a resolution is doing, it is not nothing. The study also found that self-efficacy, having the skills to change, and readiness to change — all measured before January 1 — predicted who would succeed.

That is a better foundation than despair. Resolutions are not the problem; the expectations around them are.

How long until it actually becomes automatic?

A median of 66 days — with a range from 18 to 254. That range is the most useful number in habit research and almost nobody quotes it.

The study (Lally et al., European Journal of Social Psychology, 2010) had 96 volunteers pick a daily eating, drinking or activity behaviour, perform it in a consistent context — "after breakfast", say — and rate its automaticity every day for 12 weeks. Fitting a curve to each person's scores gave the time to reach 95% of their own plateau.

Two implications follow, and they matter more than the headline:

• 21 days is fiction. So, roughly, is any single number. The median is 66, but half of people took longer, and some took the better part of a year.
• If you are on day 70 and it still feels effortful, you are inside the normal range. Not failing. The usual advice implies something has gone wrong by then, and for a large fraction of people nothing has.

Does breaking the streak ruin it?

No — and this is the finding most likely to save a resolution in February. The same study reports that missing one opportunity to perform the behaviour did not materially affect the habit formation process.

This matters because streak-breaking is where resolutions actually die, and they die from the reaction rather than the miss. One skipped run becomes evidence that the whole project has failed, and the abandonment follows the interpretation, not the event.

The data says the interpretation is wrong. A missed day is a missed day. The curve does not reset.

If you use a streak-counting app, it is worth knowing that its central motivating device is modelling something the research does not support — and that the despair it produces when the number goes back to zero is a design choice, not a fact about your brain.

What this means for reminders

Set them for longer than you think, and make them survive a miss.

• Plan for three months, not three weeks. If the median is 66 days, a reminder that you quietly stop after a fortnight is being switched off in the middle of the work.
• Attach it to a context, not just a clock. The study's participants tied behaviours to existing anchors — after breakfast, after the commute. A reminder that says "after your morning coffee" is doing something a bare 7:30am alarm is not.
• Do not make the reminder a streak counter. The message should prompt the behaviour, not report on your record. "Time for your walk" survives a bad week; "day 14 of 14" does not survive day 15.
• Reinstate rather than restart. If a fortnight goes missing, the correct move is to resume, not to begin again from a fresh January.

In ReminderIt you can set one by message — "remind me every weekday at 6pm to do my 20 minutes" — and it reads the schedule back before saving. A call suits this better than a notification for the specific reason that a notification is dismissible by reflex, and the behaviour you are trying to build is the opposite of a reflex.

What the evidence does not say

That any of this is easy, or that a prompt is sufficient. Both studies above describe what happened to groups of people; neither tells you what will happen to you, and the Lally sample was 96 volunteers choosing simple daily behaviours, not people quitting smoking.

It is also worth being straight about a common move in this genre: quoting popular authors as though they were the research. BJ Fogg is a genuine researcher; several of the other names routinely cited alongside him are writers summarising other people's work, which is a useful thing to do and not the same as having done the study. Where this page previously blurred that, it no longer does.

Sources: Norcross JC, Mrykalo MS, Blagys MD. Auld lang syne: success predictors, change processes, and self-reported outcomes of New Year's resolvers and nonresolvers, Journal of Clinical Psychology (2002). Lally P, van Jaarsveld CHM, Potts HWW, Wardle J. How are habits formed: Modelling habit formation in the real world, European Journal of Social Psychology (2010), 40(6):998–1009.