In diligence on a growth-stage software company, I would ask first for the deals it lost in the last two quarters.
A company-provided reference list arrives selected, and the customers on it can show where the product works. The lost deals show where it stops working and whether anyone inside the company has named the pattern.
I want the opportunities that got far enough to be real: a proposal priced, a security review opened, a pilot scoped, or a procurement process started. For each one, I want the recorded loss reason and then a call with the buyer. The gap between the CRM field and the buyer’s account is the finding. The CRM may list price while the buyer describes a six-month implementation against a March deadline, or an integration that needed a data engineer the buyer did not have. If that turns up in four calls out of eight, discount authority will not fix the problem, and adding sellers will reproduce the loss faster.
The installed base deserves the same treatment. A retention bridge forces churned accounts into view but makes flat renewals easy to miss. An account that renewed twice at the same seat count and with the same modules has said something specific: it kept the product without expanding its commercial footprint. A blended net revenue retention number hides how many of those accounts exist and whether they cluster in one segment or under one implementation partner.
How the company produces the list is part of the answer. If the loss reason is optional free text filled in by whoever remembers to close the opportunity, the company has not built a usable record of why it loses. Loss analysis still depends on memory.
What I want from the calls fits on half a page: two or three loss reasons I expect to repeat when the sales team is three times larger, and, where relevant, what fixing each one costs in engineering or delivery weeks. If one answer is a roadmap item, the memo should carry the ship date the company would have to hit and the quarter I would expect win rate in that segment to move. Then I find out I was wrong before the next round rather than during it.