A post-mortem on the last attempt
You’re about to start again. Before you do, answer one question: how did the last one end? Not why you think it ended in general terms — what happened on the specific days at the end of it.
Almost nobody can answer. The ending is the least memorable part of the whole episode, because there was no ending event, just a fortnight of nothing followed by the realisation. And an attempt whose failure was never diagnosed gets repeated with the same fault in it.
Why the ending is invisible
Nothing happens when a routine stops. There’s no decision, no moment, no conversation. You skip a day for a reason that was fine, then another, then the third one you notice and feel bad about, and then the noticing stops too. The event you’d need to remember is a non-event.
The story you have is the wrong story. Ask most people why they stopped and you get a summary judgement — got busy, lost motivation, life got in the way. Those are categories, not causes, and none of them names anything you could change. “Got busy” in particular is almost never literally true; the routine was a few minutes and the weeks in question contained many idle hours.
Guilt corrupts the record. If the remembered version is “I stopped because I’m bad at this,” the investigation ends there, because that conclusion offers nothing to examine. It’s also the least likely explanation, given that the same person maintains a dozen other daily behaviours without effort.
You’ve already replaced the memory with the plan. By the time you’re restarting, the interesting material has been overwritten by enthusiasm for the new version. This is why the post-mortem has to happen before, not during.
What to actually reconstruct
You need the last two weeks it was alive, in as much concrete detail as you can get. Not feelings — events. Where you were, what your days looked like, what changed in them.
Three sources are usually available, and they’re better than memory:
Whatever record you kept. If you had marks anywhere, the shape of the last month is the whole story: whether it stopped abruptly or thinned out gradually. Those are two different diagnoses.
Your calendar for that period. What was in it. New job, someone visiting, a trip, a project, a term starting, someone in the house at a different time. Cue failures are almost always visible here and almost never remembered, because a schedule change doesn’t feel like a routine change at the time.
Where the equipment ended up. If the mat got put away, when, and why — someone came over, you tidied, you moved a room around. Putting the thing away is frequently the actual end of the routine and it registers as housekeeping.
The four endings, and how to tell them apart
Nearly every ending is one of four, and each has a different fix. This matters because the default response — try harder, be more consistent — is useless for all four.
It never started properly. You did it a handful of times, always by deliberate intention, and it never attached to anything. Look for: no consistent time, sessions scattered across the day. Fix: a cue, before anything else.
It ran on novelty and stopped when that did. Two or three good weeks, dense and enthusiastic, then a fairly abrupt end with no external cause. Look for: the ending lands in the second half of the first month and nothing in your calendar explains it. Fix: build the structure while you still have the enthusiasm to build it.
A day changed shape and took the cue with it. It was going well, then something in your week moved, and the routine went quiet within days of that. Look for: a datable external change within a week of the last session. Fix: re-cue deliberately after any schedule change, and expect to.
It got too big to do on a normal day. It survived a while and thinned out slowly rather than stopping — five days a week became three became one. Look for: gradual decay, and a current version noticeably longer than the original. Fix: a floor, and check the length against what you first wrote down.
THE TEST — dating the ending
Find the last day you definitely did it. Then:
· Anything change in your week within 7 days of it?
→ cue failure. Most common,
least remembered.
· Was it dense right up to the end? → novelty ran out
· Did it thin out over a month? → it had grown too long
· Was it never dense at all? → it never had a trigger
· Did something hurt? → not an adherence
question at allTwo of these can be true at once. If so, fix the cue first; it’s cheaper and the other one is sometimes downstream of it.
Doing this without it becoming self-criticism
The post-mortem has an obvious failure mode, which is that it turns into an inventory of personal shortcomings and then you don’t restart at all.
The guard against that is to keep the inquiry strictly mechanical. You’re looking for the fault in a design, in the way you’d look for the reason a door won’t shut. “What was the routine asking me to supply on the day it stopped, and why wasn’t it available?” is answerable. “Why am I like this” is not, and doesn’t lead anywhere you can act.
The useful stance: a routine that ended was a routine that needed something you didn’t have. That’s a specification error. The specification was written by you, but it was written on a good day by someone with poor information about your worst ones, which is a normal thing to have got wrong.
What this costs
It delays the restart by half an hour, at the exact moment when the appetite to begin is highest and you’d rather just begin. That’s a real cost — the appetite is a resource and it doesn’t refill quickly — so keep the exercise short and then start the same day.
It also risks telling you something you’d rather not know, most often that the ending was predictable and that the new plan contains the identical fault.
What this doesn’t fix
Knowing why the last one died doesn’t make the next one survive. It removes one specific way of failing, which leaves the others intact.
And if there have been four attempts with four different endings, the honest reading is that you have several faults rather than one, and fixing them all in the next attempt is its own way of overloading it.