How to Keep a Promise to Yourself (When You've Broken Every One So Far)

You don't have a discipline problem. You have a structure problem.

Think about the last promise you made to another person and kept — returning something you borrowed, showing up to help a friend move. Now think about the last promise you made to yourself and broke — the gym streak, the "no phone in bed," the side project. The difference usually isn't how much you cared. It's that the first promise had a witness, a cost for breaking it, and someone who'd remember. The second had none of those.

Promises to yourself fail because you are simultaneously the person making the promise, the person it's made to, and the only judge of whether it counts. Every part of the enforcement system is you — and you're a lenient judge in your own case.

Why self-promises break: the missing three parts

People have known for millennia what makes a promise binding: a witness, stakes, and a record. An oath sworn before God, in public, in fixed words, is the maximal version. A silent 11pm resolution to "be better tomorrow" is the minimal one — and it fails on all three counts:

• No witness. Nobody else knows the promise exists, so breaking it costs you nothing socially.
• No stakes. The only penalty is private disappointment, which fades by lunch.
• No record. Memory is an accomplice. "I never really committed to that" is always available.

Everything that follows is a way of restoring one or more of those parts.

1. Say it in words a judge could rule on

"I'll work out more" is unbreakable — not because you'll keep it, but because no one, including you, can say precisely when you broke it. Vague promises are unfalsifiable, and unfalsifiable promises are worthless.

The research name for the fix is implementation intentions — Peter Gollwitzer's term for promises in the form "I will do X at time Y in place Z." His 2006 meta-analysis with Paschal Sheeran, covering 94 studies and over 8,000 participants, found a medium-to-large effect (d = .65) on goal attainment from this one change in wording. In one well-known 2002 study by Milne, Orbell and Sheeran, 91% of people who wrote down exactly when and where they would exercise followed through, against 39% in the motivation-only group.

The rule of thumb: phrase it so a stranger could rule "kept" or "broken." "Gym, Monday–Friday, before 7am, this month" is rulable. "Get in shape" is not.

2. Make it out loud, to an actual person

The witness is the single highest-leverage part to restore. The American Society of Training and Development's much-cited commitment study found the probability of completing a goal rose to 65% when people committed to another person, and around 95% with regular check-in appointments — versus roughly 10% for merely having an idea.

Treat single-study numbers with suspicion, but the direction agrees with everything else in the literature on social accountability: humans keep public promises at dramatically higher rates than private ones, because reputation is on the line. This is also why spoken commitments carry more weight than typed ones — a voice is harder to disown than a note nobody saw you write.

Practically: pick one person. Tell them the exact promise (see rule 1) and the exact date they should ask you about it. A promise with a witness and a deadline is a different object than a hope.

3. Put something real behind it

Economists call these commitment devices: arrangements where breaking the promise costs you something you set up in advance, when you were still sober-minded about your goal. The evidence is solid:

• In a Philippine study by Ashraf, Karlan and Yin (2006), smokers who put money in a forfeit-if-you-fail savings account were significantly more likely to pass a nicotine test six months later.
• Platforms like stickK, built on this research by Yale economists, report markedly higher success on contracts with financial stakes than without — especially when the forfeit goes to a cause the user dislikes (an "anti-charity").

Stakes don't have to be money. The oldest version is reputational: swearing in front of people whose opinion you care about, on something you're not willing to cheapen. What matters is that the cost is decided in advance and out of your hands in the moment of weakness.

4. Keep a record you can't quietly edit

A promise that lives in memory is negotiable forever. A promise with a timestamp is not.

This is the humble genius of things like signed contracts, public pledges, and — at the everyday level — a written log. Records do two jobs. They kill the "I never really committed" escape hatch, and they build the other asset: a visible history of kept promises. Albert Bandura's decades of work on self-efficacy show that the strongest driver of believing you can do something is direct evidence you've done it before — what he called mastery experiences. Every kept promise you can point to makes the next one more likely; a track record is a flywheel.

Make the record slightly uncomfortable to fake. A journal you can rewrite is weaker than a message sent to a friend, which is weaker than a recording of your own voice saying the words with a date on it.

5. Promise small enough to be certain, then compound

Broken self-promises aren't neutral — each one is evidence, in Bandura's sense, that your word to yourself means nothing, which makes the next break easier. So the streak matters more than the size.

The fix: scale the promise down until you'd bet a month's rent on keeping it, and keep the number of active promises to one or two. James Clear's popularization of identity-based habits gets the direction right — every kept promise is "a vote for the person you want to become" — and it's why one kept ten-minute promise beats an abandoned two-hour plan. You're not training the habit; you're training the credibility of your own word.

What this looks like in practice

Here's the whole method in one example. Instead of "I'm going to stop doomscrolling," it becomes:

"No phone in the bedroom on weeknights, for the next 30 days. I'm telling Amir today, he checks on me every Sunday, and if I break it I donate $100 to a cause I can't stand. Day one is tonight."

Rulable words. A witness. Stakes. A record with a date. That promise might still break — but it can no longer dissolve, which is how most self-promises actually die.

If you want the ritual version — saying it out loud, sealed with your voice as the permanent record, with a friend to call the verdict — that's exactly the mechanism we built On God around, and there are other accountability tools worth comparing depending on whether money, coaching, or social pressure works better on you.

Why is it so much harder to keep promises to yourself than to others?

Because you're the promisor, the promisee, and the judge, all at once. Breaking a promise to someone else costs reputation and trust; breaking one to yourself has no external cost, no witness, and a judge (you) who's motivated to acquit. Restore an outside witness and real stakes and the gap closes fast.

Should I tell people my goals or keep them secret?

Tell people the commitment and the check-in date, not just the dream. Derek Sivers' famous 'keep your goals to yourself' advice draws on Gollwitzer's identity-goal studies, where announcing an aspiration ('I'm going to be a doctor') gave a premature sense of completeness. But that research is about identity claims — the accountability literature consistently favors sharing concrete, checkable commitments with someone who will actually follow up.

What if I break the promise anyway?

Record the break honestly, take the penalty you set, and make the next promise smaller. What kills self-trust isn't a broken promise on the record — it's the quiet pattern of pretending the promise never happened. A visible 'broken' that cost you something is still evidence that your promises are real objects with real consequences.