The month revenue was real and the business was not

£61,400

That was the best month the company had ever had. Real invoices, real bank transfers, no accounting cleverness, no annual prepayments booked as monthly. I remember screenshotting the dashboard. I remember telling my co-founder we had finally turned the corner.

Eleven months later we were selling the company at a price that would have embarrassed us to say out loud in that month.

Nothing about the £61,400 was false. Every pound arrived. The problem was that revenue is an output of things that happened one to nine months earlier, and I was reading it as a statement about the present. It was a photograph of a train that had already left.

What the number was actually made of

When I finally pulled the month apart, line by line, it looked like this.

• £24,000 from two customers, both signed fourteen months prior, both on contracts that renewed the following March
• £19,600 from one-off implementation fees on deals closed in the previous two quarters
• £11,200 from the actual recurring base, which was 84 accounts
• £6,600 from a partner referral channel run by a single person who was, at that point, three weeks from resigning

So of a record month, about eighteen per cent was recurring revenue from the product we were supposedly building. The rest was history and services. Both of the large customers were later found to be using roughly nine per cent of the seats they were paying for, which is the loudest possible warning sound if you know to listen for it.

Revenue tells you what you sold. It does not tell you whether anybody wanted it.

The four numbers I check first now

I do not have a clever dashboard. I have four figures I calculate by hand at the end of each month, in a spreadsheet, because the friction of doing it manually stops me from skimming.

1. Recurring revenue from cohorts under 90 days old. New money from new customers, isolated. In that record month it was £3,100, down from £5,400 four months earlier. This is the only number that tells you what the business is doing now. Everything else is a trailing echo. If this is falling while total revenue rises, you are harvesting, not growing, and harvesting has a floor you will hit.

2. Revenue concentration at the top three accounts. Simple percentage. Above 30% you do not have a business, you have three relationships and some software. We were at 46% and I had persuaded myself this was a sign of enterprise traction. Enterprise traction is when the fourth and fifth logos look like the first three. One of ours did not.

3. Usage-weighted revenue. For each account, multiply monthly revenue by the fraction of purchased seats that were active in the last 30 days. Sum it. The gap between that and actual revenue is your quiet churn, already committed, not yet realised. Ours was £61,400 real against roughly £38,000 usage-weighted — a 38% gap. That gap arrived as churn eight and eleven months later, almost exactly on schedule, because renewal is where usage finally gets priced.

4. Gross margin on the last five deals, individually. Not blended, not average. Each one. Blended margin is where implementation heroics go to hide. We were selling at a nominal 78% margin and delivering the last five deals at 31%, 44%, 12%, 51% and 38% once you counted the engineering days spent on bespoke imports. The 12% deal was our largest logo and I had celebrated it in an all-hands.

Why the self-deception is structural, not personal

I want to resist the tidy conclusion that I was simply not being honest with myself, because that framing makes it sound like a character flaw with an easy fix. It is more mechanical than that.

Revenue is the single number every external party asks for. Investors ask it, your parents ask a version of it, your own team hears it in the all-hands. It is the only metric that is simultaneously easy to compute, socially legible, and slow-moving enough to feel stable. So it becomes the number you defend, and once you are defending a number you have stopped measuring with it.

The second structural problem is lag. In a business with a three-month sales cycle and annual contracts, the revenue line in month N is mostly determined by activity in months N-3 to N-14. You can stop doing all the work that generates revenue and watch revenue rise for two quarters. That is not a hypothetical. It is roughly what we did — I pulled two of three people off outbound to service the large accounts, and revenue went up for five consecutive months while the pipeline went to nearly nothing.

The third is that services revenue and product revenue feel identical in the bank. They are entirely different assets. One compounds and one is you selling hours at a markup with extra steps. Both look like £1 in Xero.

The honest-numbers ritual

I now do this on the first working day of each month and it takes about 40 minutes.

1. Compute the four numbers above, by hand, before opening any dashboard.
2. Write one sentence for each, in plain English, predicting what it will be next month and why.
3. Read last month's predictions before writing this month's.

Step three is the one that does the work. Keeping a written record of your own forecasts is unpleasant in a way that graphs are not, because you cannot re-narrate a sentence you wrote in March. In my first six months of doing this I was directionally wrong about new-cohort revenue four times out of six, always optimistic. That fact — that I was systematically optimistic by roughly 30% — was more useful than any individual number.

One more thing I would add if you sell to businesses: put a named renewal date and a named champion against every account over 10% of revenue, and note the date you last spoke to that champion. In the record month, our biggest customer's champion had left the company nine weeks earlier and nobody at our end had noticed. That single unrecorded fact was worth about £14,000 a month and we found it in a LinkedIn notification.

Where this leaves the good months

I am not arguing that you should refuse to enjoy a record month. Enjoy it for an evening. But treat the revenue line as a lagging report on decisions you can no longer change, and treat the four numbers as the actual news.

The uncomfortable version of this: it is entirely possible for a business to be shrinking for a year while every number a founder looks at daily goes up. That is not a rare pathology. It is the default behaviour of any business with contracts longer than its reporting cycle.

Do this before Friday

Take your last completed month. Calculate the four numbers by hand — new-cohort recurring revenue, top-three concentration, usage-weighted revenue, and per-deal margin on the last five. Do not build a dashboard. Use a spreadsheet and forty minutes.

Then write down the one number that surprised you. In my experience, roughly two-thirds of founders who do this find the surprise is number three, and it is always larger than they expected.