There's expansion revenue in your numbers that nobody expanded.
It isn't fraud and nobody chose it. It's a period-alignment artifact, it's in almost every cohort comparison I've looked at, and it takes an afternoon to find.
The Mechanism
Cohort comparisons work by measuring a base period against a later one. Call the base 2025 and the comparison 2026.
Now put a customer in it who signed in November 2025 at $2,000 a month.
Their contribution to the 2025 baseline is two months. $4,000.
Their contribution to the 2026 comparison is twelve months. $24,000.
The difference is $20,000, and it lands in the expansion bucket.
That customer bought nothing additional. They were present for the whole of one period and a sixth of the other.
The number is real revenue. It just isn't expansion, and calling it expansion means you've credited a motion that never ran.
Why It Survives
Because every part of it is individually correct. The base is right, the comparison is right, the subtraction is right. Nothing in the calculation is wrong, so nothing in the calculation flags it.
And it's invisible in aggregate. One account with a partial base period disappears into a portfolio number, which is the same reason a blended average hides the segments that behave nothing alike. You only see it by going down to the account level and asking a question nobody asks: how many months of this customer's life are in each side of this comparison?
The Part That Should Bother You
The contamination scales with acquisition.
Sign nobody new in the base period and there's none of this at all. Sign steadily and a slice of every year's expansion number is period alignment. Have a strong second half and the following year's expansion figure is substantially manufactured by the calendar.
The faster you grew, the more of your expansion number isn't expansion.
Which inverts what the number is doing. It's supposed to tell you whether the base is compounding. Instead it partly restates how many customers you acquired, in a metric explicitly designed to exclude that.
Net revenue retention excludes new logos from the calculation precisely so it measures the base rather than the funnel. A partial base period smuggles the funnel back in through the side door.
The Check
For every account in the comparison, confirm the base period and the comparison period cover the same number of months of that customer's life.
Anything that doesn't, handle one of two ways. Exclude the account from the comparison entirely, which is cleanest and shrinks your sample. Or annualize the base contribution, which keeps the account and introduces an assumption you should write down.
Then rerun the number.
It comes back lower. It also comes back usable, because what's left is customers who were fully present in both periods, which is the only population where a difference means somebody decided something.
What to Do With the Corrected Number
Two things, and the second is the one worth the afternoon.
The first is that you now know your real expansion rate, which is probably below what you've been reporting and above zero. Fine. Report the real one.
The second is that the gap between the two numbers is a measurement of how much you've been crediting to a motion that doesn't exist. If your reported expansion was $2 million and the corrected figure is $1.2 million, then $800,000 of what you thought was a working expansion engine was the calendar, and every decision you made on the strength of that engine was made on a number that wasn't describing it.
That's the same failure as a renewal that comes back higher while the customer sheds commitments. Revenue moved. Nobody bought anything. The reporting can't tell the difference, so somebody has to.
The Question Underneath All of It
Every version of this comes back to one test, and it's the same test at the account level and at the portfolio level.
Did a customer commit to something they weren't committed to before?
If the answer is no, whatever moved isn't expansion, however the spreadsheet has it filed.
Lincoln Murphy formally named and popularized Customer Success starting in 2010 and has spent 15 years connecting it to expansion revenue and commercial outcomes. Read The Premise.