Plenty of companies compute CAC payback on new-logo revenue only, and they do it on purpose. The reasoning's clean. Acquisition spend bought the customer, it didn't buy the expansion, so letting expansion into the denominator flatters the acquisition machine for work it never did.
That's a defensible position and it rests on a premise nobody says out loud.
It assumes expansion is exogenous. Something that happens to you rather than something you cause.
If that's true, excluding it is correct. Expansion arrives when it arrives, it has no relationship to who you acquired, and letting it into an efficiency number is measurement noise dressed as performance.
That premise is worth holding up to the light, because a lot of standard practice rests on it without saying so.
Most companies have earned it honestly. Expansion isn't a motion there. Nobody owns it, nothing triggers it, no forecast contains it, and what shows up shows up. Under those conditions the exclusion is right and the people doing it aren't being lazy, they're being careful.
What breaks if it isn't true
Expansion is partly a function of who you acquired.
Whether the account's got anywhere to grow into. Whether they landed in a segment with an ascension path or one that tops out. Whether the deal was written so that anything remained to sell them. Whether they were sold something they could actually use in quarter one, or handed the whole platform at a discount and left to sit on most of it.
Every one of those got decided during acquisition, and every one shows up later as expansion or the absence of it.
Which means a customer who expands at month three isn't luck. That's frequently a better acquisition, and it's better in a way the new-logo-only number is built to hide.
The cost of hiding it
Say one segment pays back in 15 months on new-logo revenue alone, and 11 months once their expansion lands. Another segment pays back in 15 both ways, because nothing ever gets added.
Those aren't the same customer and you shouldn't be spending the same amount to acquire them. You can afford to pay more for the first one. You can afford to target harder, tolerate a longer cycle, and lose more deals chasing them, because the money comes back four months sooner.
Run new-logo-only and both segments read identically. The exclusion didn't protect the metric from noise. It deleted the signal that tells you where to spend.
So run both
Payback on new-logo revenue tells you what acquisition bought.
Payback including expansion tells you what the account became.
Neither one is the honest number by itself. The gap between them is the honest number, and it measures something no single figure does: what everything after the sale is actually worth, per segment, in months.
A wide gap means your post-sale motion is producing real money and your acquisition efficiency looks worse than it is. A gap of zero means either you have no expansion motion or you acquired accounts with nowhere to go, and those two have very different fixes.
The reason this is worth the trouble
You can't decide who to go acquire more of using a number that's been cleaned of the thing that made some of them worth acquiring.
It's a reasonable answer to a real problem. It just answers it by removing the evidence instead of splitting it.
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.