Half an hour, on a fixed date

Review your numbers on a fixed date - the first Sunday of the month, half an hour, five numbers, one change - whether the month was good or bad. Most people in this trade review when a week goes badly, which guarantees every conclusion is drawn from the least representative fortnight available.

The five numbers

Nothing else belongs in the review.

Jobs completed. Take-home after every deduction, not headline fees. Hours spent, including admin and messages. Enquiries received, converted and not. Average take-home per job.

Take-home divided by hours is your real hourly rate, and it is the only figure that compares this month to last month honestly. Headline fees do not, because the deductions between a buyer's payment and your bank account move independently of your price - the four separate ones are itemised in the post on platform fees.

The enquiry count is the number that most people do not have, and it is the one that distinguishes a demand problem from a conversion problem. How to capture it costs about fifteen seconds an enquiry and is set out in the piece on tracking non-converting enquiries.

What to ignore

Any single week. Any single review. The best month, when setting expectations, and the worst month, when deciding whether to continue.

Compare against a rolling three months rather than against the immediately preceding one. Month-on-month in an irregular trade is mostly noise, and reading noise as signal produces changes that then have to be undone. Instinct does not correct this: in Bishop, Thompson and Parker (2022, Royal Society Open Science), participants trained on sample size still did no better on quiz items testing whether they neglected it. The general form of that mistake - small samples read as trends - is set out plainly in Measure My Cock's material on data, and it applies to a booking log exactly as it applies to anything else measured a few times.

What triggers a change

Three months of movement in one direction, not one month.

Effective hourly rising with flat jobs means your speed improved, and it is the cue to consider a rise - timing and size are covered in the post on the first rate rise.

Effective hourly falling with rising jobs means your mix has drifted cheap, and the fix is the rate card rather than more work.

Conversion falling with flat enquiries means the offer or the price, not visibility.

Everything flat is usually capacity, which is a different diagnosis with a different response - what a plateau actually means covers it.

Change one thing. Two changes produce a result you cannot attribute, and attribution is the entire point of having done the arithmetic.

Keep the file boring

One row per month in one sheet, kept for years. The value of this record is almost entirely in its second year, when you can see a season repeat rather than guessing whether November was unusual.

Resist scoring yourself. The numbers are inputs to one decision a month, not a verdict on the work, and treating a quiet month as a verdict is the specific error described in the post on quiet months.

Two things this review deliberately does not measure. It does not measure the quality of your assessments, which no self-report can - what a score can and cannot reliably capture is the subject of another property rather than this one. Nor does it measure anything about the tooling on the buyer's side, including automated scoring, which changes the market you price into but not the arithmetic of your own month.

For the platform's own monthly figures, whatever Rate Cock reports to judges sits on the judges page, and it is a useful cross-check against your own sheet rather than a replacement for it.