I occasionally get asked some version of “How long do I need to practice my habit before it will stick?” The coachee has often heard some magic number – 21 days, 6 weeks, 9 weeks – that will make a new habit automatic and robust. (I hear surprising different numbers, giving how similar the rest of the statement is.)

It’s an alluring idea. Habits form the backbone of our default actions. We probably would be happier, healthier, and wealthier if we could just make doing the “right thing” automatic.

But I think this mindset is missing something important about habits really work.

Oh, it’s not entirely wrong. We seem to learn cue->response patterns that we execute by default when we’re not deliberately choosing how to act. Changing those responses does reduce the need to effortfully intervene.

Yet, most habits weren’t chosen over a search space of one option. You didn’t pick up an afternoon cookie habit because that was the only food you’ve ever eaten for an afternoon snack. You picked up the afternoon cookie habit because the cookie tastes good!

It seems so obvious when you think about it. Of course we form habits we find rewarding! By default, we gravitate towards actions that feel good in the moment (or at the very least, not bad). Even maladaptive habits usually provide immediate pleasure or relief, like procrastinating on dealing with an email for weeks. The overall feeling of dread and lateness might be building, but each moment you can avoid dealing with the issue provides fleeting relief from facing it.

I don’t have a precise answer to “How rewarding does it need to be?” Roughly, it seems like the more immediately rewarding, the easier it is to form a habit. Like, I had a really easy time developing an afternoon ice-cream habit. I had to hunt more for the reward of daily Pilates (which for me is learning to enjoy the activity a bit, pride at keeping my streak going, and “reward cuddles” from my supportive partner).

Chapter 2 of Effortless by Greg McKeown has some good tips/examples of finding immediately enjoyable aspects of tasks. He mostly focuses on pairing clearly enjoyable actions with the necessary tasks, but I liked one exception example: “I noticed an element of the [weekly financial] meeting I’d previously overlooked: the whole exercise involved tidying up a messy area, the family finances! Just noticing that creating order out of chaos was a part of the experience gave me a tangible and immediate benefit.”

Repetition can still help. You still need to retrain the cue->response pattern. (Even just knowing what you want to do is a helpful step.) But forcing yourself to repeat a disliked action daily is usually not sufficient to make a robust habit. I’ve tried using stickk.com to make myself do habits to get over this hump, but the habit never lasted long (though sometimes the financial commitment was worth doing for other reasons). Having to use financial penalties to force myself is a good sign the habit isn’t rewarding on its own.  

“But what about the 6-weeks thing? I think there’s research on it, and I kept doing one habit after practicing it for six weeks!”

The habits that make it to six weeks are the ones that you can do consistently for six weeks. I’m guessing >95% of attempted habits never make it to six weeks of daily practice, and that “habits that make it to six weeks (or whatever) will last” is just a selection effect. I expect that finding a rewarding habit to begin with is going to make it a lot easier to reach that milestone.

And the research commonly cited doesn’t really support the conclusion.

According to random internet posts, the 21-day number came from plastic surgeon’s observation in a pop-sci book that it took at least 21 days for patients to get used to their new face. (I didn’t investigate this claim, but a claim in a pop-sci book being overblown by the internet wouldn’t surprise me.)

When I googled “how long does it take to form a habit,” eight of the nine posts on the first cited Lally (2009) (and the ninth post didn’t cite any studies). The usual take away is that the study found that it took between 18 and 254 days for participants to form a new habit. …Which is a bit surprising, since the study only ran for 84 days. Quickly skimming the math, the 254 days sounds like it came from an extrapolation based on fitting an asymptotic curve model. But the authors might have fitted the model based on half the data points (the ones that fit well) and then…just omitted the other half? I’m still confused.

More importantly, even if all the math checks out when read carefully by someone who knows about study methodology, I think the real-world takeaways are limited. Digging into the study, “how long to form a new habit” is actually “how long before self-reports of how automatic the habit felt plateaued.” So, it tells us that behaviors you do repeatedly will start to feel more automatic over time. (And then probably stop feeling increasingly automatic past some variable threshold for repetitions.)

This is some support for the idea that repeating a behavior makes it a habit, but I’m hesitant to conclude that it supports the inference that you just need to repeat an action enough times and voila, you have a robust habit. It doesn’t address the people who dropped out of the study. Nor does it tell us if the habits will continue after the study.

After talking with dozens of people about their habits, I can tell you that “automatic” habits still fail. Just go on a trip, get sick, or have a new baby. (Though it’s often easier to reboot the lapsed habits than to start entirely new habits.)

Although, if you can force yourself to repeat an action every day for 254 days, maybe you have enough willpower to just muscle through. If that’s you, feel free to ignore this post. For the rest of us, I recommend looking for new habits that you find rewarding.