Comparing Periods Fairly
Five things that make two periods incomparable, all of them present in most long records, and how to account for each.
The main use of a long record is comparing yourself against yourself. Doing it carelessly produces confident conclusions from artefacts.
A concrete way to test the method is outlined in the full explanation. Use a short pilot and keep only the fields that still make sense when comparing the record months later.
1. Coverage
Count recorded days, not calendar days.
When this method depends on versioned project records, the material from GitHub offers a helpful independent point of comparison.
A year with 340 recorded days compared against one with 280 will show a decline that is entirely about recording.
Divide by coverage or state it. This is the commonest error and the easiest to fix.
2. Definition changes
Check the definitions file for anything dated between the two periods.
A category that widened, a rounding rule that changed, a granularity that coarsened — each makes a like-for-like comparison invalid without adjustment.
Where a change falls between the periods, compare the parts that did not change or accept the comparison is indicative rather than precise.
3. Tool changes
A migration almost always introduces a discontinuity: different rounding, different handling of breaks, different day boundaries.
If a tool change falls between your periods, expect a step in the data that is not a step in your life.
This is why the migration date belongs in the definitions file.
4. Life changes
A new job, a move, a child, an illness.
These are the things you are usually trying to measure, and they also change what you record and how.
Note them in the record at the time, because in three years you will see the step and not remember the cause.
5. Season
Covered in its own note: compare the same period across years, not adjacent periods.
A quarter against the previous quarter is almost always a seasonal comparison dressed as a trend.
The fair comparison, assembled
Same calendar period. Adjusted for coverage. No definition or tool change between. Life changes noted.
Where all four hold, the comparison means something.
Where they do not, say which and read the result accordingly rather than abandoning the comparison entirely.
Stating uncertainty to yourself
Write the caveat down when you record the finding.
"Q1 2026 up 12% on Q1 2025, but coverage was better and the category widened in March" is a useful note.
"Q1 up 12%" is a figure you will repeat in two years having forgotten the qualification, which is how personal data becomes personal mythology.
The comparison nobody should make
Against other people's figures.
Nobody collects the same way, the definitions differ, and published figures come from people whose data flatters them.
Your record is comparable with itself and with nothing else, which is a limitation and also the entire point.
What to check
Does your analysis divide by recorded days?
Do you check the definitions file before comparing periods?
Are tool changes and life changes dated in the record?
And when you last quoted a figure about your own year, did the caveat survive with it?
The point
Same calendar period, adjusted for coverage, no definition or tool change between, life changes noted.
Where all four hold the comparison means something.
Worth adding
Never compare against other people's figures.
Nobody collects the same way, and published figures come from people whose data flatters them.
Additionally
Write the caveat down when you record the finding.
A figure repeated in two years without its qualification is how personal data becomes personal mythology.
Finally
Count recorded days rather than calendar days.
A year with 340 recorded compared against one with 280 shows a decline that is entirely about recording.
In summary
Five things make periods incomparable: coverage, definition changes, tool changes, life changes and season.
Most long records contain all five.
Last word
The practice this points at is unglamorous: a small daily act, a short weekly correction, an annual read, and a written page of conventions beside the data. None of it is difficult and all of it is easy to skip, which is why most records do not reach the point where they start paying.
More in this section
These notes are about keeping a personal record for years rather than for a fortnight: durability beats accuracy, and a rough record kept for five years answers questions a perfect record of one month cannot. Tool guides are included as practical comparisons, while the underlying method remains independent of any single product. Nothing here is legal, tax or medical advice: requirements differ by jurisdiction and situation.