21 Days

Why is 21 the magic number? Do a quick search for anything on habit forming and you’ll find that the ‘experts’ agree that it takes exactly three weeks to form a habit. If you can do something for 21 days in a row, it’s likely going to stick with you for life.

The origin of the 21 day rule is unclear, although other sites have attributed it to a surgeon working with amputation in the 1960s. I’ve been working on a habit-formation guide for my online clients and through my search for some empirical research on habit formation I found that it may take a little bit longer than 21 days to get those habits set in stone. Here’s one study that took a look at how long it takes for certain behaviours to become automatic.

Can we get university students to eat healthy, drink water and exercise?

Lally et al (1) recruited 96 university students (almost 2/3 female) to participate in a habit-setting exercise. Participants were required to choose one habit (related to either healthy eating, drinking or exercise) and attempt to maintain it for 84 days. The habit couldn’t be one that they already do and had to be performed once per day in response to a cue. Throughout the study they were required to log in to a website daily to report whether they had completed the task or not and complete a 12-item questionnaire relating to the ‘automaticity’ of the task.

The authors created this study to test the assertion that habits were formed through a relationship of diminishing returns, described by the authors as an ‘asymptotic’ relationship. Under this model, early repetitions of the behaviour contribute heavily to whether or not the task becomes automatic for the individual, however as repetitions continue past some point, each additional rep contributes less and less to habit formation until a plateau in the relationship occurs.

Automaticity Graph

Did they stick to it?

Out of the 96 participants, 39 tracked habits that could fit the curve expected by the authors. Across all three behaviours the median time to the plateau in automaticity was 66 days, with a range of 18 to 254 days (predicted), for the most part much longer than the 21 days we frequently see online.

There was no significant difference between habit behaviours for both time to reach the plateau in automaticity and the actual automaticity score at this plateau, however compliance was highest for the simplest activity (water drinking) as compared with either eating healthy foods or completing daily exercise.

To see if missing days altered the degree of automaticity for the chosen habit, the authors selected all periods where a behaviour was missed on one day, but followed by three successful days. The degree of automaticity of the behaviour wasn’t affected, so it seems missing a single day won’t be a problem as long as you get yourself right back on track.

What about everyone else?

Like any good study, I’m left with more questions than answers although it definitely doesn’t help that the social sciences aren’t my forte.

Perhaps the biggest question is of the 57 participants who failed to ingrain new habits or at least fit the curvilinear relationship expected of the researchers. Are these a population of habit-resistant people, and if so, how can we identify them and what strategies can we use to help them? Based on the numbers enrolled in the study, the people that didn’t fit the relationship outnumber the ones who did, I’m sure I’m not the only one who’d like to know a little more about them.

As alluded to by the researchers, there’s certainly a logical argument for habit formation taking longer for complex as opposed to simple behaviours. In the exercise world, is someone more likely to stick to a simple running or walking program than a multi-day strength training protocol? To add another element, most of us are able to string together a series of movements that resemble running, but many would be considered naive to most weight-training exercises. I’m sure the novelty of many strength exercises could act as a hindrance in the early stages of habit formation, as this added complexity could increase the time required to form a successful habit.

And finally, when considering missed days of the habit, how many missed attempts will result in decreased automaticity? I’m sure no one would argue that immediately getting back on track is best, but where’s the threshold, two days, three days or more? At what point do we accept that the habit may not be realistic for the individual at this point in time, and that the number of missed days and reduced automaticity indicate an unsustainable habit.

Ripped calendar days

Should we start selling 66 days to a new you?

When considering this in the context of the ‘rule of 21’ championed on the internet, I’m left wondering what the role of a coach would be in developing these habits. The presence of a trained coach would likely accelerate early changes in automaticity, and their active monitoring and intervention may make the rule of 21 more of a standard than when people are left to their own devices, as seen in the study above. I’ve certainly focused on 21 day intervals in the past for clients with success, however the primary focus was more on formation of solid behaviours than on a number of days.

At the end of the day, when it comes to forming new habits it takes as long as it takes. We need to shift the focus away from seemingly arbitrary numbers and realize that habits will take various amounts of time to form based on the individual and the complexity of the chosen behaviours. Instead, an approach that focuses on simple tasks and building up and tracking a pattern of success is one that will be sustainable and contribute most to behaviour change.


    Aashish Sharma says:

    I really like the scientific approach applied here. Thanks for posting it here for everyone to see! I was toying with the idea of trying out the 21-day rule myself when I did a quick search about it and found this page. I agree that commitment to the 'activity/habit' matters more than the number of days itself. I'll give it a try and see how it works for me.
    I'd like to ask you permission to post a link (no content) to this page on my own blog. Please respond to my e-mail address to let me know if I have your consent or not.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Share on Facebook
Share on Twitter
Subscribe to Newsletter

Dan Ogborn