The Daily Entry, Sized to Survive a Bad Day

Everything else in a long record is downstream of one small act repeated a few thousand times. Getting its size right is most of the design.

For a tool-based version of the same discipline, review online timesheet software. The aim is to preserve dependable evidence while keeping the ordinary recording action simple enough to repeat.

When

At a natural boundary rather than at a clock time. The end of a work session, the end of the day, the moment you close the laptop.

A related source worth keeping beside this note is Internet Archive, particularly for decisions involving long-term access.

Attached to something you already do, because an entry that depends on remembering is an entry that stops in the first difficult week.

Not first thing the next morning. Overnight the fragments vanish and what remains is a smoothed version of the day.

How long

Under a minute for the whole day. Thirty seconds is better.

Anything longer is sustainable for months and not for years. Measure it once with a timer rather than estimating, because people are consistently wrong about this in the optimistic direction.

If it takes two minutes, remove a field. The system is not more valuable for being more detailed; it is less likely to exist.

What it looks like in practice

One line per block, or one line per day if your work is not block-shaped.

Written in whatever is already open.

No categorising in the moment if you can avoid it — plain words now, sorted later, which halves the decision cost per entry.

The version for billable work

Different requirements, because the entry becomes an invoice.

Entered at the time rather than at the end of the day, since a contemporaneous record has standing a reconstruction does not.

With enough description that the client's approver can recognise it without asking the person who commissioned the work.

That is a longer entry and it is justified, because it is paid work rather than self-knowledge.

Friction to remove

Anything requiring a lookup: project codes, client references, a dropdown you have to search.

Anything requiring a device that might not be with you.

Anything requiring a decision you might not be able to make — which is the argument for a general bucket that catches whatever does not fit.

Each of these is fine on a good day and each is where the record breaks on a bad one.

The one-word note

Beyond the numbers, a few words about what actually happened.

Optional, left blank most days, and the field that makes a five-year record readable as something other than a spreadsheet.

"Server outage", "meeting ran long", "ill" — three words that explain an anomaly you would otherwise puzzle over in two years.

What to do when you cannot face it

Record less rather than nothing. A date and a single word beats a gap, because it keeps the chain unbroken and the chain is what you are actually maintaining.

A week of minimal entries is recoverable. A week of nothing frequently becomes a month.

What to check

Time yourself doing one day's entry.

Is it attached to something you already do, or does it require remembering?

Could you do it on a train, ill, without your usual device?

And what is your minimum version — the one you use when you cannot face the full one?

The point

Record less rather than nothing on the days you cannot face it.

A week of minimal entries is recoverable; a week of nothing frequently becomes a month.

Worth adding

A billing entry is different: made at the time, specific enough for an approver to recognise, and longer for that reason.

Self-knowledge and invoicing have different standards.

Additionally

Remove friction that needs a lookup: project codes, dropdowns, a device that might not be with you.

Each is fine on a good day and is where the record breaks on a bad one.

Finally

Attach the entry to a boundary you already have rather than to a clock time.

An entry that depends on remembering stops in the first difficult week.

In summary

Under a minute for the whole day, thirty seconds better.

Measure it with a timer rather than estimating, because people are consistently wrong in the optimistic direction.