Tracking that doesn't become a second job
The spreadsheet lasted eleven days. It had a column for each component, a column for how it felt, a column for notes, and by day six the notes column was empty, by day nine the feel column was empty, and by day twelve the whole file was closed and hasn’t been opened since.
Roughly everyone has a version of this. What’s worth noticing is the order things failed in: the tracking collapsed before the routine did, and then took the routine with it, because the tracking had become how you knew whether you’d done it.
That’s the specific risk with measurement. Not that it doesn’t help — it does — but that it’s a second behaviour with its own adherence problem, and you’ve now got two things to keep going instead of one.
What tracking is actually for
Three distinct jobs, usually conflated, with very different requirements:
Knowing whether you did it today. Trivial in principle, and it’s the one that matters most. Your memory for “did I do this today” is worse than you think, particularly for short routines attached to other activities. A mark somewhere answers it in a second.
Seeing the trend over weeks. Whether this is holding up or quietly decaying. Almost nobody notices decay in real time — routines don’t stop, they thin out — and the trend is the only way to see it while it’s still cheap to fix.
Recording detail about the sessions themselves. What you did, how it felt, what was tight, what progressed. This is the expensive one, and it’s the one people build systems for.
The third job is the only one that requires anything more than a mark, and it’s the one with the weakest case. Detail is useful if you’ll act on it — a physio who wants to know, a specific problem you’re troubleshooting, a progression you’re managing. If you’re not going to act on it, you’re doing data entry to no purpose and paying for it every day.
Judge a method by its cost per day, not its capability
Everything about tracking follows from one asymmetry: you pay the cost every day and you collect the benefit occasionally. A method that takes ninety seconds costs you about nine hours a year, and you’ll look at the output maybe a dozen times.
So the question to ask of any method is not “what can this record” but “what does this cost me on the worst day I’ll have.” Which is the same question as the one for the routine itself, for the same reason.
TRACKING COST — the test
On your worst realistic day, after finishing the routine:
· Marking it takes under 5 seconds and no decisions
→ sustainable
· You'd have to open something, navigate,
and type → it will be skipped,
then abandoned
· There's a field you'd have to think
about how to fill in → that field is why
it will stop
· You'd tell yourself you'll log it later → you won't, and the
gap will make the
record uselessThe “log it later” row is the one that kills spreadsheets specifically. Retroactive logging produces an incomplete record, an incomplete record stops being trustworthy, and an untrustworthy record isn’t worth maintaining. The whole thing then unwinds in about a week.
Methods, ranked by cost
A mark on a wall calendar. Paper, pen, one stroke. Effectively zero cost, visible without being opened, gives you today and the trend simultaneously because a month of marks is a shape you can read at a glance. It cannot record detail, which is a feature. For most people this is the correct answer and it loses to apps only on the grounds of being unglamorous.
A single mark in whatever you already use. If a calendar or notes app is already open every day, one character in it costs almost nothing. The condition is already open — a tool you’d have to launch for this purpose is not this category.
A count, no dates. A tally you increment. Cheapest possible, and it never resets, which dodges the streak cliff entirely. You lose the ability to see patterns, since there are no dates.
A weekly tick against a target. Five out of seven, marked as you go. Slightly more structure, and it makes a missed day a non-event rather than a break. Good balance.
Anything with fields. A spreadsheet, a form, a structured log. High cost, high capability, and only justified when something specific is being troubleshot. Treat it as temporary by default: run it for the two weeks you need the data and then stop deliberately, rather than letting it die and taking the routine with it.
The trap: the record becomes the goal
Once there’s a record, the record starts to exert pressure of its own, and the pressure isn’t always pointed the right way.
Two ways this goes wrong.
The chart becomes the objective. Days start getting done to keep the line intact rather than because they’re useful, which sounds harmless and mostly is — until the day you’re ill and doing it anyway to protect a graph. If the record is making you do something on a day you shouldn’t, the record has taken over.
Marking becomes doing. Rarer, real, and mostly happens with granular logs: the session gets shortened while the log entry stays the same shape, because the entry is what you interact with. The map replaces the territory quietly.
Both are arguments for coarse tracking. A single mark can’t be gamed in any interesting way, because there’s nothing in it to game.
What to do with the data
Almost nothing, most of the time. Look at it once a month, for about a minute, asking one question: is the density the same as it was?
Not “did I hit the target.” Density — whether last month has visibly more marks than this one. Routines usually don’t stop; they thin. Five days a week becomes four, then three, then a fortnight where you’d have sworn you were keeping it up. A wall calendar shows this instantly and memory does not.
If the density is dropping, that’s a design finding, and the usual suspects are that the cue changed or the routine grew.
What this costs
Coarse tracking means you lose the detail. If in six months you want to know what you were doing when your shoulder felt best, a wall of ticks won’t tell you. That’s a genuine loss and for a small number of people — managing a specific rehab, working with a clinician who wants specifics — the detailed log is worth its cost.
For everyone else the trade is straightforward: you give up information you were never going to use in exchange for a measurement system that still exists in March.
What this doesn’t fix
Tracking tells you whether it’s happening. It has no influence on whether it happens — that’s the cue and the size of the thing. A perfect record of not doing something is still a record of not doing it.
And no tracking method has ever survived a routine that was too long. It just documents the decline in more detail.