Renewal risk board

This is a churn early-warning radar for the Customer Success team at Rippling. It groups 14 existing customers by the quarter their contract renews, bands each one red, amber or green, and never shows the band on its own: every point is attached to a dated event with the sentence Trayo used to justify the match, the source link and the people involved. Four checks run against each account — the champion or exec sponsor leaving, hiring that implies building or buying a replacement, news of layoffs/restructuring/acquisition, and a new leader arriving over the HR, IT or Finance function we sell into. Expanding an account opens the full evidence trail and a ‘who to talk to’ tab with the standing owner of that function, their verified work email, and a brief on their track record so the intro is not cold.

Everything on the board came out of the Trayo API today. Fourteen accounts were imported (two needed their identity corrected with a LinkedIn handle before discovery would match them). Three new signals were created — champion-departure, replacement-build-or-buy, downsizing-or-ownership-change — and run alongside two the workspace already had, people-ops-joiner and it-finance-exec-joiner. Four discovery runs settled: a 180-day backfill over 7 accounts (86 source results, 3 events), a 365-day deep pass at a raised result budget (154 results, 0 new — the backfill had already caught everything), a 365-day pass over 8 newly added accounts (83 results, 4 events), and the weekly run at a 10-day lookback over all 14 (34 results, exactly 1 new event). That last run is the no-double-count proof: it re-read work the backfill had seen and Trayo deduplicated all of it but the one genuinely new story, dated the previous day. Stakeholder search returned 73 people across the accounts against the workspace’s own buyer personas; 20 email lookups returned 19 verified work addresses; 10 person-research reports supplied the career history and inferred pressures in the briefs.

Three things are worth saying plainly. First, the champion-left check was built as a job_change signal through discovery, as specified — and that source reports people arriving at an account, not leaving it. Across 14 accounts and a 365-day lookback it returned no named departure, so the lane fires on an external hire landing over our function (the seat changed hands) and the app says in as many words that Trayo never named the predecessor. Second, replacement-build-or-buy matched nothing at all: a real empty result, not a skipped check. Third, the relationship history the brief would ideally lead with — meetings, QBRs, the last email thread — lives in a CRM that is not connected, so that panel is absent rather than invented, and Push to Salesforce, Draft intro and Post to Slack are previews that toast what would have been sent. The customer list itself stands in for an upload: the companies are real Rippling customers taken from the workspace ICP and Rippling’s published customer stories, but renewal quarter, ARR, seat count and CSM name are not API concepts and were written for this build. One imported account, Clariti, was dropped and removed from the workspace — Trayo resolved a domain for it but returned no stakeholders, so its identity could not be trusted. The palette applied cleanly: #7A005D became the primary and #FFFFFF the page, with the library noting that a hueless white canvas takes a soft brand tint (#fef8fc) and that, with no shell colour supplied, the top bar uses a deep step of the primary (#750459).