Every review you run on a sales call is looking for what went wrong.
Loss post-mortems. Churn analysis. Call coaching. All of it is failure analysis, and failure analysis is genuinely useful. I've written the failure side of this myself.
Here's a question nobody runs.
Give me thirty customers you have a sales call recording for, who later bought more from you.
Both Answers Are Worth Having
That's the mark of a question worth asking. There's no response to it that leaves you where you started.
If they produce the thirty, you have something you've never had. The positive pattern: what got said on a call that preceded a customer choosing to buy more. That's the only analysis that can tell you what to repeat. Everything else tells you what to stop.
If they can't produce them, that's the finding.
It isn't a record-keeping problem, which is how it'll get explained in the room. It means no sales conversation in the company has ever been connected to a later expansion. Either it isn't happening, or it's happening and nobody has looked. Both are expensive. The second one is worse, because the information was sitting right there.
Removing the Bad Doesn't Produce the Good
This is what makes failure-only analysis structurally incomplete rather than merely partial.
You can eliminate every value-limiting move from your sales process. No discounting to close, no cancel-anytime, nothing promised you can't reliably deliver, nothing stuffed into the bundle to make the number look bigger. Do all of it perfectly.
You'll have a sales process that's stopped destroying lifetime value. You won't have one that builds it.
Those are different jobs, and no amount of the first produces the second.
What You're Looking For
Concretely, on a call that preceded an expansion, at least one of these happened.
Somebody named a future state the customer was working toward, specifically enough that you'd know later whether they got there. That's a milestone, and it's what an expansion attaches to.
Somebody said what comes after that, and roughly when. That's the whole orchestration motion, and it takes about fifteen seconds of a sales call.
Somebody left something out on purpose and said so, which means there's now a thing that exists, has a price, and isn't included yet.
Or none of that happened and the customer expanded anyway. Also worth knowing, because it means the demand was strong enough to survive your process rather than to be produced by it. That's a ceiling you can raise.
The Asymmetry
Most companies can hand you their churn reasons by category. Somebody built that taxonomy, somebody maintains it, and it gets reported upward.
Ask the same company for their expansion reasons by category and you get a pause.
That gap isn't an accident of tooling. It's what happens when one motion has a loss attached and the other doesn't. Churn hurts, so it got instrumented. Expansion that never happens costs nothing anybody can point at, so the category list was never built.
Ask It Anyway
Thirty customers. Sales call recording. Bought more later.
If that list exists, read it before you read another loss post-mortem. It's the more useful document and nobody has opened it.
If it doesn't exist, you just learned something that costs more than anything in the churn pile.
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.