Evidence and interview

What to ask to see for each dimension, and how to score what comes back. The rule throughout: a capability that cannot be shown is scored as absent.

Evidence requests by dimension

Dimension	Ask to see	Scores as not met when
ICP definition	The written ICP document, and the last list built from it	The ICP lives in someone's head, or the last list did not follow it
Lead scoring	The scoring model: inputs, weights, and when it was last changed	Nobody can say what the inputs are, or it has never been recalibrated
Outreach	A live sequence, with its steps and channels	Reps write from scratch, or "multi-channel" means email plus the occasional call
Data quality	A duplicate count and a field completeness count from the CRM, plus the written data protocol	The counts cannot be produced, or the protocol exists but no cleanup has run
Feedback loops	Notes or outputs from the last two optimization reviews, and the last A/B test result	Reviews are described but not recorded, or no test has concluded
Automation	The list of live workflows, and which ones require human approval	Workflows exist that nobody owns, or none have an approval step

Interview questions

Ask these of the revenue leader and, separately, of one frontline seller. Where the two answers differ, trust the seller's.

1. Walk me through how an account goes from unknown to worked. Where does it come from, who decides it is worth working, and what happens first?
2. What made you work the last three accounts you touched?
3. When did the scoring model last change, and what changed it?
4. What did the last optimization review decide, and what happened because of it?
5. Which parts of outreach run without a person starting them? Which of those does a person check first?
6. If I pulled every account in the CRM for your top segment, how many would be duplicates?
7. What is the forecast accuracy for the last two quarters?

Scoring the answers

• Intended versus running. A process that is planned, being rolled out, or "mostly followed" is scored as not met. Score the process that ran last month.
• Leader versus seller. A gap between the two accounts is itself a finding. It usually places the true level one below the leader's description.
• Artifact age. An ICP document or scoring model untouched through a major change in product, pricing or market is scored one level lower than its contents suggest, because it no longer describes the business.

Safety

The diagnostic is read-only. Pull counts and view records; never edit, merge, delete or re-score anything in the CRM while assessing it. Every change the diagnostic recommends goes to a named owner for approval.