The default assumption is that more frequent is better. It is not. Beyond a certain point, extra observations add noise rather than signal, take up attention, and — for some people — make the thing being measured worse.
The right frequency depends on what you are trying to find out.
Frequency follows the question
Is something changing over months? Weekly is plenty. Daily data adds day-to-day noise that obscures the trend you are looking for.
Did a treatment change do something? Daily for the first two or three weeks, then back off. This is the one situation where daily is clearly right: the effect is new, the timing matters, and the window is short.
Is there a pattern within the day? Two or three times a day, for a week or two. Then stop — you are looking for a shape, and once you have it, further measurement tells you nothing new.
Is there a pattern across a month? Daily, briefly, for two or three cycles. The pattern you are looking for is periodic, so you need coverage of the period.
Is this getting worse in a way I should raise? Instances as they happen, with dates. Not a schedule at all.
Nothing in particular? Then do not track. This is the case that produces most of the harm, and it is the most common.
Why daily forever is usually wrong
Noise dominates. Day-to-day variation in mood, clarity and energy is large and mostly meaningless. A daily series is mostly the noise, and reading it invites finding meaning in fluctuation.
Adherence decays, and decays unevenly. People log more on bad days, which biases everything: a record kept sporadically is a record of your worst days, and it will look like a decline.
It changes what is measured. Rating your own attention requires attending to your attention. Rating your mood requires attending to your mood. Both are known to make things feel worse, and the effect is not small.
It makes you a patient full-time. The most serious cost and the hardest to see. A daily practice of assessing your own health keeps a health identity permanently in the foreground, and for most conditions that is not neutral.
The general shape that works
Intensive, briefly, for a specific question. Then stop. Light and regular, for a trend. Weekly, briefly. Event-based, always. Note things when they happen, with a date.
Most people's best arrangement is: a weekly line, plus notes when something happens, plus a two-week intensive period around any change in treatment.
Signs it has stopped earning its place
Worth checking against honestly:
- You cannot say what question you are answering
- You feel worse after logging
- You are checking whether today is a good day rather than noticing
- You reread old entries to compare
- You log more on bad days than good ones
- You are recording things you would not have noticed before you started
- Nobody has ever looked at it, including you
- It has been going for months with no decision made from it
Any two of those and it is worth stopping for a month. Stopping is a legitimate option and the ability to stop is what distinguishes tracking from monitoring.
When to stop entirely
When the question is answered. You found the afternoon pattern. Write down the finding, put the log away.
When it is making you worse. Non-negotiable, and it is worth saying to a clinician rather than quietly abandoning — because "I stopped because it was making me anxious" is itself useful clinical information.
When things are stable. A stable condition does not need a daily record. Notes when something changes are enough.
When nothing is ever done with it. If six months of data has not informed a single decision, the data was not the missing ingredient.
The one exception worth protecting
Keep the ability to record instances, always. Not a schedule — a place to write something down when it happens.
That is the highest-value form of tracking and the lowest-cost. A dated line when something notable occurs costs nothing, requires no routine, creates no monitoring habit, and produces exactly what a clinician needs: specific, dated, contemporaneous observations.
If you only do one thing, do that one.
Before a change in treatment
One argument for tracking more, and it is a good one: the data collected before a change is the only thing that makes the change measurable.
Two weeks of baseline before starting something is worth more than two months after. Almost nobody has it, because the two weeks before starting are the two weeks you are waiting to start. If you know a change is coming, that is the time to write things down — and then you have something to compare against.