The situation
An illustrative B2B SaaS company in Austin — a workflow automation tool serving operations teams — was at $80k MRR with a 4.2% monthly churn rate. At that rate, they were churning roughly $3,400 of MRR every month.
The founding team knew the product was strong. Support ticket sentiment was positive. NPS scores were decent. But customers were cancelling anyway — and by the time the founder found out, the decision had been made 6-8 weeks earlier.
The CS function was one person reacting to cancellation requests. There was no proactive monitoring, no health score, and no systematic outreach to at-risk accounts.
What got shipped
The snapshot went live on day 5 after purchase. The team focused on three workflows from the SaaS Snapshot launch guide:
1. Behavioral health score calculation. Product event webhooks were connected in the first week (login frequency, workflow execution count, active seat utilization, support ticket age). The health score workflow ran for the first time on day 8 and immediately flagged 11 accounts in the red zone (score below 40) and 23 in the yellow zone (score 40-65).
2. At-risk alert → CS task workflow. When a health score dropped below 40, a high-priority task was created for the CS rep with the account’s behavioral data attached. The CS rep had a full picture of what had changed before making the first call.
3. Four-touch dunning workflow. Stripe webhooks were connected on day 3. The dunning sequence went live on day 6. In the first 30 days, 8 contacts entered the dunning workflow and 5 recovered (recovery rate: 62.5%).
Illustrative outcomes
By day 90:
- Monthly churn rate dropped from 4.2% to 2.1% — not entirely from the health score work (product improvements were happening in parallel) but health-score-driven CS interventions saved 7 accounts in the first 60 days that the team could specifically attribute to the alert workflow.
- Failed-payment recovery added approximately $3,400/month in recovered MRR that previously churned silently.
- NRR improved 8 percentage points, from 94% to 102% — expansion revenue workflows contributed as seat utilization alerts drove 4 upgrade conversations in the period.
- CS response time on at-risk accounts dropped from an average of 4 hours (when someone manually noticed an at-risk signal) to 45 minutes (automated alert + task creation gave the CS rep a structured prompt to act immediately).
What worked
The health score alerts changed the CS team’s posture from reactive to proactive. The CS rep’s summary: “Before, I found out about at-risk accounts when they sent a cancellation email. Now I get an alert 6 weeks before that happens. The conversations are completely different — customers are surprised that we noticed and reached out proactively.”
The dunning workflow was the immediate revenue win. The team had been recovering roughly 0% of failed payments (no systematic outreach). Moving to a 60%+ recovery rate on a consistent basis added meaningful MRR that required no acquisition spend.
What would be done differently
If this scenario ran again, the product event webhook connection would be prioritized on day 1 rather than day 5. Every day the health score runs without accurate behavioral data is a day the at-risk alerts are less precise. The sooner the webhooks are connected, the sooner the health scores are reliable.
The NPS survey would also be configured before going live (rather than in week 2 as happened in this scenario). The first NPS promoter responses came in during week 3 — had the review request workflow been ready, those responses would have generated G2 reviews. The reviews came eventually, but the timing gap cost a few early promoter responses.
Caveat
This is an illustrative scenario. Actual results depend on product quality, CS team execution, customer segment, and market conditions. SaaS churn is affected by many variables beyond automation — including product-market fit, pricing, and competitive dynamics. Your mileage will differ.
“We'd been talking about building proper churn alerting for over a year. The snapshot gave us a working version in a week — and the health scores immediately flagged three accounts we would have lost without intervention.”