--- title: "How to Stop Starting Over on the Same Habit" description: "Count the restarts before you plan the next one. Why the next attempt fails for the same structural reason as all the others, and what to change first." source_url: "https://thecommitapp.com/blog/starting-over-again/" updated: 2026-08-14 --- # How to stop *starting over* on the same habit You already know which goal this is about — you thought of it before the end of that sentence. Before you plan the next attempt, count the last few. People who restart a lot are not short of plans. Ask how to stop starting over and you get another one: start smaller, be kinder to yourself. The last attempt had all of that. ## First, do the arithmetic Three questions, about one specific goal rather than goals in general. Answer them honestly and the diagnosis falls out on its own. 1. **How many times have you started this one?** Count January. Count the Monday after the trip. Count the app you have installed four times. If you lost count, write that down instead — it is the more useful answer. 2. **How long did the best run last?** Most people can name it to the day, and name the week it fell over. That precision is the tell. It is a pattern, not bad luck. 3. **What was actually different each time?** Usually the plan. Better schedule, smaller unit, cleaner start date. Now ask what was different about the moment you did not feel like it — that is where every attempt ended, and it was identical every time. > Every attempt had a better plan than the last one. None of them had anything holding it up. ## The count never moves because the mechanism never changes Every restart ran on the same machinery: you, deciding again, on your own, at the worst hour of the day. Nothing checked whether you actually did it. Nobody noticed when you stopped, and skipping cost nothing. That is three missing parts, and not one of them is discipline. - **Doesn't hold · What every attempt had.** A better plan than the last one. A start date. Your own word as the only thing holding it up — unchecked, unwitnessed, free to walk away from at 6am. - **Holds · What none of them had.** Evidence you could not fake. A person who could see whether you showed up. Terms fixed in advance by the version of you who still cared. ## Build the next one differently, not better - **A · Decide the evidence up front.** Not a checkbox you award yourself. A photo, a check-in at the place, a Health reading, a screen-time number — chosen when you write the commitment, not when it is due. Captured live, so “close enough” has nowhere to go. - **B · Name who sees it.** One person is enough. Quitting in private is easy, which is the whole reason it keeps working on you. Quitting in front of someone who watched you sign up for it is a different experience. - **C · Fix the terms while you are clear-headed.** Decide in advance what a miss costs: nothing but the streak, the apps you chose blocked for a day, screen time you earn back with pushups, or money. Pick the smallest one that would change your Tuesday. ## Then decide now what a miss means This is the part that decides whether there is another restart. A miss is one day. Starting over is a decision you make about that day, usually within an hour of it. - Come back today, not Monday. A future start date is a week of permission. - Come back at a quarter of the size. This attempt does not need to be the most ambitious one. - Do not make up the session you missed. Doubling up is how one bad day becomes a bad week. - Track the rate, not the run. Twenty-nine days out of thirty reads as zero on a streak counter and as twenty-nine on a rate. Deciding in advance is a small effect, and a real one. The trial behind that claim, and what it did not show, is set out in what a commitment device is (https://thecommitapp.com/commitment-device/). ## Common questions ### How many restarts is too many? No number disqualifies you. The count is a diagnostic: if it keeps rising while your plans keep getting better, the plan was never the variable. ### Should I wait until Monday to start again? No. A start date in the future is a week of permission to keep not doing it. Do a token version today and let Monday be day four. ### Does one missed day mean I have to start over? Only if you decide it does. Missing one day rarely breaks a habit — what breaks it is treating the counter going to zero as the end of the attempt. Commit gives a 30-minute grace window past the deadline, then several hours more, plus grace tokens that mark a day skipped rather than missed. ### Why would someone watching change anything? Because wanting it was never the missing piece — you have wanted it every time. What was missing was anyone who would notice the difference between showing up and not. ## Sources - The commitment-contract evidence this piece points to is set out, with the citation, at https://thecommitapp.com/commitment-device/ - Commit feature claims come from a read-only audit of the Commit iOS beta, August 2026. If something here is out of date, tell us through the contact page and we will correct it. Sources checked on 14 August 2026. If a number here is wrong, or a paper says something other than what we report, write to support@thecommitapp.com and we will correct it. ### Commit is in public beta Commit is that mechanism in one app: write the commitment in a sentence, choose the evidence, name who sees it, set what a miss costs, and sign. iPhone only, free in public beta through TestFlight and not on the App Store. Prefer the small print first? Read the Commitment Agreement, which is the actual contract you would sign, or start with what a commitment device is. ## Keep reading - [What one missed day actually costs](https://thecommitapp.com/blog/missed-one-day/) - [Why you go hard for two weeks then stop](https://thecommitapp.com/blog/how-to-stay-consistent/) - [How a commitment device works](https://thecommitapp.com/commitment-device/) - [The nine kinds of proof](https://thecommitapp.com/how-it-works/)