A churn play every CSM would actually run
An enterprise book was losing accounts to causes nobody was catching early enough, and the playbooks that existed were too heavy for anyone to follow under pressure.
- Client
- Aulysius (monday.com Platinum Partner)
- Role
- Customer Success Manager
- Period
- Jul 2024 to Jul 2025
- Scope
- Enterprise and mid-tier book
01 / The problem
Churn was being diagnosed after the funeral
Accounts were leaving, and by the time anyone recognised an account was at risk the customer had usually already decided internally. Post-mortems produced explanations, not prevention.
There were playbooks. Nobody ran them. They were long, they assumed time nobody had, and they asked CSMs to act on instinct rather than on a signal, so under pressure people fell back on whatever they personally thought worked.
02 / Context
Capable people, no shared method
Aulysius is a monday.com Platinum Partner, which means customers arrive with implementation expectations as well as software. I managed strategic enterprise and mid-tier accounts across the portfolio, onboarding through renewal and expansion, and acted as escalation point when accounts went critical.
The team had good CSMs. What it didn't have was one method they all used, which meant risk was visible to whoever happened to be looking and invisible to everyone else.
03 / Approach
Treat churn as a failure mode with a signature
My background is QHSE engineering, where the job is to find the conditions that precede an incident rather than explain the incident afterwards. Same sequence here.
- Find the failure modes. Work backwards from accounts already lost to the conditions they shared before anyone noticed.
- Make them detectable. Map those conditions onto signals you actually have (platform usage, log data, engagement patterns, stakeholder changes).
- Design the intervention. One defined response per failure mode, scoped so a CSM could run it inside a normal week.
- Make it the default. A play only works if it's easier to run than to improvise.
A playbook nobody runs isn't a playbook. It's a document.
04 / What I built
One play, adopted across the whole team
- Usage-based risk detection. Monitoring platform and log data to surface friction before the customer raised it.
- Structured success planning. Plans tied to each customer's own objectives, not to our feature list.
- Reviews that carry ROI. Every QBR and EBR answered "what did this return" in the customer's numbers.
- A defined escalation path. At-risk accounts routed to support and product against an agreed standard.
- ROI reporting. Translating feature adoption into a figure a finance team recognises.
05 / Results
Two quarters later
The play outlasted my involvement in individual accounts, which was the point. A method that depends on one person isn't a method.