Every resource on page one for this query is a scorecard review. A documented SOP has the reviewer confirm the score's inputs are fresh, check the weights against an approved model, then classify flagged accounts by severity before the meeting. A quarterly-review skill format adds a four-component score with its own yellow-cap rule. Both assume the score itself is the hard part. Neither asks which accounts get reviewed at all, and a CSM working from a flagged list reviews the accounts that made themselves visible this week, not the ones actually declining.
Tversky and Kahneman's research on the availability heuristic found that people judge frequency by how easily instances come to mind, not by the actual rate of occurrence. A support ticket is memorable by construction, filed on purpose so someone will read it. An account whose usage has been sliding for two months with nobody complaining generates nothing to recall, so a review built on recall cannot reach it. Crossing contact load against usage trend, not just totaling a score, is what makes that account visible on purpose instead of invisible by default.
For one fictional vendor, Castellan Software, and a sixty-account book reviewed in Q2: 23 accounts, 38.3 percent, generated a ticket or a call this quarter, and a reactive review touches exactly those. Eight more, 13.3 percent, had usage quietly declining with no contact at all. One quarter later those eight had churned or downgraded at 62.5 percent against 14.3 percent for the loud-but-stable accounts, 41.7 percent of every bad outcome in the book from 13.3 percent of it. An account here with high contact and declining usage routes straight to the save play tracker instead.
Customer Health Review Process Template