Customer success automation · Pre-emptive retention · Subscription
Your health-score dashboard says “at risk” on Tuesday. By Friday it’s a cancellation. Move from a score to typed triggers.
You run retention across thousands of subscription accounts. Three support tickets in 48 hours about the same feature. The executive sponsor stopped opening emails. The customer admin removed 7 of 12 seats. All three signals collapse into one “at-risk” score on the dashboard. By the time your CSM looks Friday morning, the cancel button has already been pushed. Typed triggers per signal class assign an intervention + an owner + a deadline before the QBR.
Published May 30, 2026
Why a score is not the intervention
The health-score dashboard does its job. It says: account ABC is at risk. The CSM opens the account, sees a 47/100, and does not know which of the seven signal classes drove the drop. Did the sponsor disengage? Did a senior user leave? Was there a billing dispute? Did they evaluate a competitor? The dashboard does not say. The CSM either guesses or asks for help.
Meanwhile, the actual signal that mattered most fired at 2pm Tuesday. By the time the CSM looks Friday, the cancel decision has been made. The dashboard was right. The intervention never happened.
The gap is not your health-score platform. The gap is between the score and the action. Typed triggers per signal class close that gap by naming the signal, the intervention, the owner, and the deadline as the trigger fires.
We’ve built the trigger layer for subscription operators. Here’s what we know.
You probably already use a health-score primitive — ChurnZero, Catalyst, Vitally, Gainsight, Pendo, Totango, Custify, Profitwell Retain. Each is good at the health-score primitive. The gap is the per-signal-class trigger layer that converts each signal into a typed intervention with a routed owner and a deadline. We bring the trigger taxonomy as a starting point and tune it against your account history.
We have built this layer for subscription operators across verticals. We know which signal classes compound first (usage decay and exec-sponsor disengagement co-occur in most preventable churn). We know which triggers should route to a CSM and which should fire automated outreach. We bring the CSM-workload-balance playbook so triggers do not overwhelm your team.
How we get from score-only to typed triggers
Step 1 — Tier 1 AI Readiness Assessment ($10k, 2-3 weeks). We audit your current churn-precursor data coverage. We sample your last 90 days of churned + saved accounts and run a per-signal-class diff: which signals fired before churn, which were detected, which were detected but did not produce intervention. Output: the per-signal-class trigger spec, the per-CSM workload budget, and the per-trigger routing-rule starter.
Step 2 — Tier 2 AI Swarm Setup Sprint ($25-50k, 4-8 weeks). We build the trigger layer end-to-end: per-source signal-ingestion connectors, per-signal-class classifier, per-trigger CSM-routing wire, per-trigger automated-outreach hook, per-trigger attribution event emission, per-CSM workload-balance enforcement. Your team receives the running system, all source code, all credentials.
Step 3 — Tier 3 Fractional CMO with AI Swarm ($15-25k/month, 6-month minimum, 1-2 days/wk). We operate the layer in production. Tune per-signal-class weights as new precursors emerge. Update the per-trigger intervention message templates. Run the weekly false-positive review with your CS leadership. Roll up a monthly per-signal-class trigger-effectiveness report.
What changes for you
You stop discovering the Friday cancellation Monday morning in the QBR-prep deck. The triggers fired Tuesday and Wednesday; the interventions ran while the decision window was still open.
You stop asking your CSMs to manually triage every at-risk account. The triggers route by signal class to the right person with the right context. Your CSMs spend their time on conversations, not on dashboards.
You can answer the question your VP of CS asks every quarterly review: which signal classes produced the highest save-rate this quarter, and which CSMs converted which signal classes most reliably. The per-trigger attribution events roll up per signal class per CSM.
You can onboard a new product line or vertical without re-training the health-score model. The signal taxonomy extends; the trigger pipeline stays the same.
Frequently asked
How is a typed trigger different from a health-score threshold playbook?
A health-score threshold playbook fires when the aggregate score crosses a number. The CSM gets an "at-risk" notification with no context. A typed trigger fires on a specific signal class — support-ticket spike, seat reduction, executive-sponsor disengagement, renewal-contact departure — and carries the signal context, the assigned owner, the intervention shape, and the deadline. The score is for the dashboard. The triggers are what produce action. A score-only system can be right that an account is at risk and still produce no meaningful intervention. A typed-trigger system names the action and the owner before the QBR.
What are the canonical signal classes worth triggering on?
Seven recur across subscription operators. Usage decay (primary feature unused for 14 days when baseline is daily). Support-ticket spike (3+ tickets in 48 hours or one severity-1, especially on the same topic). Seat reduction (admin removed >25% of seats in last 30 days). Executive-sponsor disengagement (original buyer stopped opening emails, stopped joining calls, stopped logging in). Renewal-contact departure (renewal POC left the company per enriched contact data). Payment-method risk (payment method expires <30 days, or last payment failed). Competitive-evaluation signal (admin created competitor accounts + engaged with your competitor RFP-prep content). Each fires a different intervention with different urgency + owner + message template.
How do triggers split between CSM-assignment and automated outreach?
Per-signal-class routing. Executive-sponsor disengagement, renewal-contact departure, and competitive-evaluation signals route to a CSM with a deadline — these need human relationship-rebuild and cannot be automated cleanly. Usage decay and payment-method risk fire automated outreach (re-engagement email, in-app tour, payment-update workflow). Support-ticket spike and seat reduction fire both in parallel (CSM ping + executive-summary email; automated check-in if CSM does not respond within 8 hours). The routing rules are operator-side configuration; the trigger layer enforces them. The classification of human-vs-automated is your decision; the trigger layer carries it.
What does Completions commit to on Tier 3 if we run this layer in production?
Tier 3 process commitments include: signal-classification at sub-minute latency for streaming signals; per-signal-class trigger evaluation on every change-event from the underlying customer-state stream; per-CSM workload-balance enforcement (no CSM receives more than N typed triggers per day); weekly false-positive review of triggers that fired without producing intervention value; quarterly review of the per-signal-class weight tuning. We commit to the operating discipline. Per-signal recall + precision are tuned against your account corpus and recorded as engagement KPIs.
Who owns the signal catalog, the routing rules, and the playbook post-engagement?
Your team owns the per-signal-class catalog, the per-trigger CSM-routing rules, the automated-outreach templates, the per-account assignment data, and the credentials. Completions owns the orchestration knowledge: the signal-classification runbook, the per-trigger tuning history, the per-CSM workload-balance playbook. At engagement end we transition operational ownership back to your team over 30-60 days with documented handover.
How does the trigger stream connect to the rest of the retention + revenue stack?
The trigger layer subscribes upstream to the change-event-emission stream (customer-state changes), the sentiment-intent-classification urgency stream (high-urgency messages), and your product analytics (usage signals). It publishes downstream: every trigger emits a typed event into the at-risk pipeline, where the recovery-rate dashboard rolls up per-CSM + per-signal-class outcomes and the lifecycle-flow architecture fires the automated touchpoints. Five layers, one trigger contract. The trigger layer owns the contract; each consumer subscribes.
Start with the audit
Tier 1 AI Readiness Assessment ($10k, 2-3 weeks): we audit your churn-precursor data coverage, sample your last 90 days of churned + saved accounts against the signals that fired, and produce the per-signal-class trigger spec + per-CSM workload budget. If you decide to build, Tier 2 ships the layer. If you decide to operate it with us, Tier 3 runs it in production. You choose the next step at each gate.
Related reading
If you also care about what feeds the triggers or what they fire into:
- Multi-label urgency routing — the upstream layer that publishes high-urgency at-risk events the trigger pipeline subscribes to.
- Real-time data sync — the canonical change stream carrying customer-state changes the trigger layer subscribes to.
- Save-flow propensity scoring — the offer-presentation engine triggers fire automated outreach through.
- Recovery-rate dashboard — the per-CSM per-signal-class outcome rollup triggers feed into.
- Lifecycle email + SMS — the automated outreach layer triggers route into for non-CSM-assigned signals.
- Fractional CMO with AI Swarm — the Tier 3 engagement that operates the trigger layer in production.