Every app that asks you to track something promises that it will be worth it. Very few show you what “worth it” looks like, which makes the request hard to evaluate before you have already spent the year.
So here is the concrete version — what accumulates, when each thing becomes readable, and what stays out of reach no matter how complete the record is.
A note on the numbers below. The example figures in this article are constructed for illustration. They are not taken from a real person’s record, anonymised or otherwise. Nothing anyone logs in this app is visible to us — it stays on the device — which means we could not use a real one even if we thought it was appropriate, and we do not.
Months 1 to 3: it stops being a guess
Almost nothing feels like it is happening in the first month. One period start date is not a cycle length, and two is not a range.
Around the third cycle, one specific thing changes: you can state cycle length as a range instead of an average. That sounds minor and is not. “About 45 days” is a number that describes none of your actual cycles; “between 34 and 61 days over four cycles” is a finding, and it is the kind of sentence that gets written in the notes.
At the same point, symptom counts start to mean something. A count of days per month is only interpretable next to other months — one month of “fatigue on 22 days” is a number with nothing to compare it to.
What you can do with three months: walk into an appointment and answer the questions precisely instead of approximately. That is most of what an appointment can act on, and it arrives long before the year is up.
Months 4 to 6: things that move together become visible
With six months of dated entries, a second kind of question opens up: do two of these run together?
An illustrative finding, in the shape these take:
Bloating runs higher in the three days after nights under six hours — eleven times out of fourteen.
Three things about that sentence are doing work, and all three are the point:
- It has a count in it. Eleven out of fourteen, not “often”.
- It is falsifiable. If the next fourteen occurrences run four out of fourteen, it was noise, and you will be able to tell.
- It is a pattern and not a cause. The direction is not established. Poor sleep may be contributing to the bloating; the bloating may be disturbing sleep; something else may be driving both. A record shows things moving together and can never say why — anything claiming otherwise is overreaching, and it is the specific way health apps mislead people.
This is also the point at which the record starts contradicting your memory, which is uncomfortable and useful. Memory keeps the worst weeks and quietly generalises them.
Months 7 to 12: the slow things
Three categories only become readable at this range.
Lab trends. Values that repeat annually or six-monthly need at least two points, and three before a direction means anything. A testosterone of 2.1 says very little; 1.4, then 1.8, then 2.1 across eighteen months is a direction, and a direction is something to raise.
Medication effects. Anything aimed at hair growth cannot be assessed before roughly six months — the hair cycle sets a floor on how fast a change can appear. Judging it earlier is judging a period too short to contain the answer.
Cycle drift. Whether the range itself is moving. A year of cycles running 34 to 61 is a different report from a year in which they ran 45 to 61 early and 29 to 38 later, and only the second half of a long record can show that.
What comes out at the end
The output that matters is not a dashboard. It is a document someone else can read in two minutes.
For an illustrative year of roughly 250 logged days out of 365 — which is a realistic completion rate, not a perfect one — a report of this kind contains:
- Cycle history: every start date, each gap, and the range across the year
- Symptom frequency: days affected per month, per symptom, side by side
- Medications: what, and from when
- Lab results: each value with its date and its reference range, in order
Four sections, no interpretation, no verdict. The clinical reading is a clinician’s job; the record’s job is to make sure there is something to read.
The 250 out of 365 is worth dwelling on. A record with gaps is a record. The missing days are usually the weeks you felt worst, which makes them informative rather than lost — and an app that made you feel you had failed at 250 would have got 90.
The four things a year still cannot tell you
Stated plainly, because the omissions are the honest part.
Why. A record shows what happened and what moved with what. Mechanism is not in it, and no amount of additional logging puts it there.
Causation. Ever. Two things moving together is two things moving together. This is not a limitation to be overcome with more data; it is what the observation is.
A diagnosis. Not the first one and not a revision to it. That is made against criteria by someone qualified to apply them.
What to do. A record makes an appointment specific. It does not replace the appointment, and it does not contain a treatment.
Whether it is worth it
The honest answer depends on one thing: whether you expect to have appointments about this in future.
If you do, the record is the difference between ten minutes spent reconstructing six months from memory and ten minutes spent on the actual question. That difference compounds across every appointment, and the version of it that exists in a year is unreachable by any amount of effort on the morning of.
If you do not — if things are stable and you are content — then three months of cycle dates in a notes app is a perfectly reasonable amount of effort to spend, and there is no obligation to do more.
What would be a mistake is starting in the week before the appointment. Not because a fortnight is worthless — it beats a reconstruction, and reconstructions skew towards the bad days. But the useful record for March’s appointment was started in December, and the only version of that decision available today is the one about next year.