Kit Revenus Irréguliers

Because an average is a promise you cannot keep. One brilliant month pulls the mean up and leaves you short in the other eleven; a median leaves you short every second month by definition. The 25th percentile is a floor you clear in roughly nine months out of twelve — low enough to survive a quiet month, high enough to live on. The sheet shows all three next to each other and says in words why it uses the one it uses, so you can disagree with it knowingly.

It will hold your log, your tax set-aside and your client tracker from day one. It will not print a self-paycheck until you have six logged months — before that it shows a dash and the line "log at least 6 months before this number means anything". Three data points cannot tell you what a normal month looks like, and we would rather show nothing than a confident wrong number.

That advice was written for salaried people, whose main risk is one sudden job loss. Your risk is dispersion — several small shortfalls in a row. The workbook uses MIN(12, MAX(3, ROUNDUP(3 + 12 × CV, 0))) months, so a steady retainer contractor lands near three months and someone with a coefficient of variation above 0.7 lands near eleven or twelve. The base, the multiplier and both limits are ordinary yellow cells on Settings, and the rule is printed on the sheet so you can argue with it.

Yes. Upload the .xlsx to Google Drive, open it and choose File → Save as Google Sheets. Every formula uses functions Sheets supports — PERCENTILE, MEDIAN, STDEV, SUMIFS, COUNTIFS, AVERAGEIFS, INDEX/MATCH, SUMPRODUCT, LARGE, EDATE, TEXT. There are no macros, no XLOOKUP, no FILTER and no dynamic arrays, so the self-paycheck and the CV come out to the same cent in both. A one-page PDF in the download walks you through it.

No, and it never will. You type your own set-aside percentage on Settings, and the Tax Set-Aside sheet only compares what that rate says you owed with what you actually moved to the tax account. Rules differ by country, state and business type — ask your accountant or your tax authority. Nothing in this workbook is tax or financial advice.

Payment date, everywhere. The log has both, so you get average days to pay per client, but every total, average, percentile and chart counts money on the day it arrived. Smoothing an income on invoices you have not been paid for would defeat the point. Invoices with no payment date show as outstanding on the Client Tracker and nowhere else.

If you logged nothing for that month, it is left out of the window rather than counted as zero — a false zero would drag every average down and inflate your volatility. If you genuinely earned nothing but want it counted, log a zero-value row for that month and it will be included. The near-zero month in the sample data is there on purpose so you can see the buffer logic react.

Yes. Fourteen months of a fictional freelancer — thirty-seven invoices, twelve clients — sit only in the yellow input cells, so every chart works the moment you open the file. Clear Income Log A7:K606, Monthly Roll-up H7:I30, Client Tracker B7:B46 and the yellow cells on Settings. The Read me tab lists every range. Formulas, dropdowns and charts stay intact.

Seven pages, delivered four times over: US Letter and A4, colour and ink saver. A cover, a how-to page, an invoice tracker for 24 invoices with an outstanding-and-days-to-pay summary, a twelve-month tax set-aside log with a running gap, a monthly reconciliation page, a worst-month plan you fill in on a good month and read on a bad one, and an annual review. They mirror the workbook, so the numbers you copy across always mean the same thing.

If that is what the arithmetic says, yes. Step 4 of the Self-Paycheck sheet takes your average month, subtracts tax at your rate and your fixed monthly costs, and compares what is left with the paycheck the rule suggests. When there is a gap it prints the gap in cash and says the percentile rule sizes a paycheck you usually clear but does not check that your fixed costs leave room for it. Most tools quietly hand you a number instead.