Identity-based habits: become someone who trains

"Become the kind of person who trains" sounds like something printed on a water bottle. It's also one of the few pieces of pop habit advice with real research behind it. Psychologists have been measuring how strongly someone identifies as an exerciser since 1994, and that score predicts who keeps training better than almost anything else in the field.

I build Bespoke Fit, a live coaching service, so I've got a stake in you training consistently. That's why every claim below is sourced, and why I'll say plainly where the science runs thinner than the self-help version lets on.

Yes, it's real. But the causation runs backwards from how most advice sells it. You don't think your way into being an exerciser. You do a few sessions, notice you did them, and slowly start believing it. If you're inactive today, action has to come first.

The scale that measures this

Psychologists have had a way to measure this since 1994: the Exercise Identity Scale, nine short statements like "I consider myself an exerciser," rated on how true each one feels.¹ Researchers have used it for three decades to ask one question: does seeing yourself as an exerciser predict whether you actually exercise?

It does, and not by a little. In 2016, researchers pooled 62 datasets on identity and physical activity and landed on a correlation of r = .44 between the two.² Most single psychological measures in this field don't get anywhere near that. The same research found identity helps close the gap between intending to train and actually doing it, the same gap I've written about in why willpower is the wrong plan.

The theory underneath is older. In 1972, psychologist Daryl Bem argued we often figure out who we are the way an outsider would, by watching what we do.³ You don't lace up your shoes because you're an exerciser. You watch yourself lace up, again, and conclude you must be someone who does this.

If that sounds like Atomic Habits, it should. James Clear popularized this exact framing in 2018: every action is a vote for the person you're becoming. It's a good frame. Clear didn't invent the science. He packaged it, and what's underneath is older and messier than the vote metaphor lets on.

Which comes first, the identity or the training?

It depends where you're starting from.

If you're inactive right now, you don't have an identity to lean on yet. You have to build one. A 2021 randomized trial in Psychology of Sport and Exercise put 276 inactive women through 16 weeks of supervised training, and their exercise identity climbed the entire way through, from 3.27 to 4.07 on a 7-point scale, purely from showing up.⁴ The gain wasn't just a feeling: each extra point predicted roughly 16 more minutes of weekly exercise six months after the program ended.

Once you're already training regularly, the arrow points both ways. Your sense of being an exerciser also predicts whether you keep going. That's a nicer problem than the one most people reading this actually have.

Either way, you can't install the identity directly. Researchers pooled 40 effect sizes across 4,939 adults in 2025 and found the interventions moved identity itself only a little (g = 0.18); in the subset of those studies that also tracked actual behavior, it moved a lot more (g = 0.61). You can't talk yourself into being an exerciser. You have to show yourself, over and over, until it's true.

Think of identity as a running average of your recent behavior. Once it's high enough, it starts protecting the actions that built it.

Three ways to generate proof fast

If identity updates on evidence, the job is to generate evidence you can't argue with. Three things move fastest.

Pick a floor, not a goal. Choose the smallest session that still counts as training to you, something like 20 minutes twice a week, and make that your bad-week minimum. "I trained both days, even that week" is evidence. A skipped ambitious plan is evidence for the other side, which is also why long streaks backfire, as I've written about in why workout streaks fail: one miss ends a streak. It doesn't end an identity.

Say the schedule like it's already a fact. There's a real difference between "I'll try to work out this week" and "I train Tuesdays and Fridays." The first is an intention, and plenty of people mean to train and don't. The second is a claim with a timestamp on it. After three true Tuesdays, it stops being aspiration and starts being description.

Leave the evidence lying around. Shoes by the door, a bag packed the night before, a calendar block labeled "Training" instead of "gym?" None of this is decoration. It reminds a busy brain who it already decided to be, right when the decision is actually happening.

Why other people speed it up

Identity isn't only private. The Exercise Identity Scale itself asks how you describe yourself to others, not just how you feel inside. You can privately suspect you're becoming someone who trains. It lands differently once someone else already treats it as settled.

Being known by name does this automatically. The class where the instructor notices you're back. The training partner who texts "usual time?" without asking whether. Each person who expects you holds a copy of your identity, and every session you show up for confirms it from the outside.

A coach is the strongest version of this. Notice that "my coach" is already an identity claim: the sentence assumes you're the kind of person who trains seriously enough to have one. And a booked session with an actual human changes what skipping means. You're not breaking a plan anymore. You're standing someone up.

What the research can't tell you yet

Most of this evidence is correlational and self-reported, and even the researchers behind the 2016 pooled analysis rated the underlying study quality as modest. Causation runs differently depending on where someone starts, so "act first, identity follows" fits the newcomer trials, not a settled law for everyone. Interventions aimed straight at identity barely move it, and the 2025 analysis flagged some risk of publication bias sitting underneath that number too. Nobody has pinned down the minimum dose of training that makes an identity stick. Sixteen weeks is one trial's window, not a proven threshold.

Common questions

How long before this actually feels true?

Longer than the internet's 21 days. In the 16-week trial above, identity moved gradually across the whole program, not in one quick jump. Habit research backs that up: one study that tracked 96 people found new routines took 18 to 254 days to feel automatic, a median of 66.⁵ Expect months, and expect the feeling to trail the behavior, not lead it.

Am I lying if I call myself an exerciser after two sessions?

Only if you claim the vague version, which is why blanket affirmations don't do much here. Claim the verifiable version instead. "I'm someone who trains Tuesdays" is just true after two Tuesdays. The broader identity gets built from evidence like that, so make claims your own behavior already backs up.

Is this just Atomic Habits with extra steps?

The framing you've probably heard is Clear's, and it's a fair one. But the Exercise Identity Scale predates the book by 24 years, and the research adds two things the popular version tends to skip: most of the evidence is correlational, and identity follows behavior far more reliably than it can be built directly.

Does missing a week undo the identity?

No, and that's the real advantage over streak-based habits. A streak dies on one missed day. Identity behaves more like a running average. It doesn't reset over a bad week. The only real risk is the story you tell yourself about the miss, so tell the accurate one: someone who trains had a bad week, and someone who trains comes back.

The solo version, and what a coach adds

You can build this yourself. Pick your floor and write it down somewhere you'll see it. State your schedule out loud to one other person. Put your shoes where you'll trip over them. Find one person, anyone, who'll notice if you don't show, a class, a training partner, a standing lifting date. Give it 16 weeks, then check back in on how it feels.

Bespoke Fit packages the last piece: a dedicated coach who learns your name, your history, and the equipment you actually own, with unlimited live sessions already on the calendar, so "I train Tuesdays with my coach" is true in week one, not month four. See how it works. Your first month is $49, then $299 a month as a founding member, locked for life, one of 25 spots. The identity still has to come from you showing up. A coach just makes sure someone's there watching the evidence pile up.

Sources

1. Anderson, D.F. & Cychosz, C.M., 1994. Development of an exercise identity scale. Perceptual and Motor Skills. ↩

2. Rhodes, R.E., Kaushal, N. & Quinlan, A., 2016. Is physical activity a part of who I am? A review and meta-analysis of identity, schema and physical activity. Health Psychology Review. ↩

3. Bem, D.J., 1972. Self-perception theory. Advances in Experimental Social Psychology, Vol. 6. ↩

4. Gillman, A.S., Stevens, C.J. & Bryan, A.D., 2021. Women's exercise identity increases after a 16-week exercise RCT and is linked to behavior maintenance at follow-up. Psychology of Sport and Exercise. ↩

5. Lally, P., van Jaarsveld, C.H.M., Potts, H.W.W. & Wardle, J., 2010. How are habits formed: Modelling habit formation in the real world. European Journal of Social Psychology. ↩