How We Helped a B2B SaaS Company Cut Churn by 22% in Two Quarters

A $50M ARR B2B SaaS company with 400 enterprise accounts and a customer success team of 14 CSMs engaged us to solve a retention problem. Their annual churn was 8%, concentrated in accounts in the $20K–$60K ARR tier — the tier where CSMs had the heaviest book-of-business loads and the least time for proactive engagement. By the time customers signaled churn intent, it was usually too late to reverse it. The intervention needed to happen 60–90 days earlier. Their team didn't have a system that could tell them when to intervene. We built one.

The Problem With 28 Accounts Per CSM

With 14 CSMs covering 400 accounts, the math didn't allow for deep proactive engagement with every customer. CSMs prioritized by ARR — the largest accounts got the most attention. Mid-tier accounts got attention when they asked for it. "Asking for it" often meant submitting a support ticket, which was rarely the real signal. The real signal — declining usage, feature adoption regression, executive sponsor departure — lived in product data and CRM notes that nobody had time to systematically review.

QBR prep had the same problem. Each CSM spent 3–4 hours per account per quarter assembling the review deck. At 28 accounts each, that was 84–112 hours of prep per CSM per quarter — two and a half weeks of capacity going to deck assembly.

CSMs are relationship managers. If they spend two and a half weeks a quarter making decks, they're not managing relationships. They're managing PowerPoint.

What We Built

Continuous health scoring: A weekly automation pulls product usage data, support ticket volume and sentiment, NPS scores, contract value, and executive engagement signals for every account. A scoring model weights these inputs — usage decline carries the highest weight, followed by executive sponsor change and support escalation patterns — producing a health score from 0–100. Scores below 65 trigger a CSM alert. Scores below 50 trigger an escalation to the VP's dashboard and a required 30-day intervention plan.

Automated QBR prep: Five business days before each scheduled QBR, an automation assembles the account data package: usage trend by feature area (last 90 days vs. prior 90), support ticket summary, open items from the prior QBR, renewal timeline, and expansion opportunities based on usage patterns. We configured Claude to draft QBR narrative and slide content in the CSM's voice using the account data as context. CSM review time dropped from 3–4 hours to 45 minutes per account.

Churn risk escalation: When an account's health score drops more than 15 points in a 30-day period, an escalation workflow triggers: the CSM gets an alert with the score trend and contributing factors, the VP receives a summary for the weekly at-risk review, and a 14-day follow-up task is created in the CRM automatically. No at-risk account falls through the cracks because a CSM was too busy to notice.

22% Churn Reduction in Two Quarters

The health scoring system identified 47 at-risk accounts in the first month of operation — accounts that CSMs had not flagged as at risk. Of those, 31 received proactive interventions before any churn signal had been communicated by the customer. 26 renewed. The remaining 5 churned despite intervention.

Over two quarters, annual churn dropped from 8% to 6.2%. Net revenue retention improved from 104% to 112%. The QBR automation recovered approximately 1,400 CSM hours per quarter — capacity that went to proactive expansion conversations and relationship-building in the accounts most likely to grow.

"The health score doesn't tell us what to say," the VP of Customer Success told us at the quarterly review. "It tells us which conversations to have, and when. That timing is everything."

Why This Engagement Pattern Works

Health scoring and QBR automation are the two highest-impact CS investments for teams with more than 20 accounts per CSM. The health score solves the coverage problem — no CSM can proactively monitor 28 accounts manually at the signal level you need to catch churn early. The QBR automation solves the capacity problem. The data needed is almost always already available: product usage logs, support tickets, CRM history, NPS. The missing piece is the automation layer that makes that data actionable in real time, not just visible when someone has time to look.