Net revenue retention is a composite. That gets mentioned and then dropped, and dropping it's where the trouble starts, because a composite made of three things with three different causes can't be owned by one person.
Retention. Contraction. Expansion. Same number, three mechanisms, and almost nothing that drives one drives another.
Contraction
An account gets rightsized at renewal. The buyer audits what got used against what got paid for and cuts the difference.
Work backwards and you get a chain that's got nothing to do with the renewal.
Something went into the deal the customer didn't need yet. It closed the quarter, so it went in, at a blended price the buyer had no way to evaluate. Then it sat there. Adoption of that piece never happened, sometimes because nobody drove it and sometimes because the thing genuinely doesn't work well enough to adopt without a fight: the integration is fragile, the setup is long, the workflow assumes something the customer's business doesn't do.
At renewal the buyer counts. Unused capacity is the easiest line to cut in any negotiation, and the cut isn't a surprise. It got decided the day it was bundled in.
So contraction's mostly a sales-sequencing outcome with an adoption multiplier, and both inputs were fixed long before the renewal conversation.
Churn
Different mechanism entirely.
Some is fit that was never there, decided during a cycle where somebody sold into a use case the product half serves. Some is delivery. The thing was right and it never landed, which is an implementation and product question. Some is a champion leaving, which is nobody's fault and shows up as a number anyway.
Almost none of that shares a cause with contraction. Which is why treating both as one metric produces plans that address neither.
Expansion
The third one, and the only one that's a growth mechanism rather than a loss mechanism.
It fails two ways. Either there's nothing left to sell, because everything went in at signature, or there's something left and nobody asked at the moment it was worth asking. The first's a packaging decision. The second's an orchestration problem, and it's the only one of the three a post-sale motion can move on its own.
Now look at where the review happens
Three mechanisms. The causes sit in what got sold, what got built, what got implemented, what got staffed in support, and what got asked for and when.
The review happens in one place, and that place owns about one of the five inputs.
This isn't an argument that the number should be nobody's. It's that a composite reviewed as though it were simple produces a plan against whichever input the reviewer actually controls, which is almost never the one that was binding. The quarter goes to adoption because adoption is available, in a year where the binding constraint was that everything got bundled at signature and there was nothing left to sell.
Then next quarter it's still soft, and now there's a performance conversation instead of a packaging conversation.
What to do instead
Split the number before you review it. Not as an analysis exercise. As the agenda.
Three lines: what we lost to accounts leaving, what we lost to accounts shrinking, what we gained from accounts growing. Each with its own owner, each measured on the input that owner decides.
Then ask the only question that matters at that meeting. For the contraction line, what was sold that never got used. For the churn line, what was promised that never arrived. For the expansion line, what was ready and never got offered.
Every one of those is answerable. Not one of them is answerable by whoever's currently presenting the slide.
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.