← Archive

Case #014

Customers were signaling churn 40 days before they left. Email engagement dropping. Site visits shortening. Cart abandonment rising. The data was there. Nobody was watching it.

Churn PredictionEngagement DecayLeading vs Lagging

The 40-Day Churn Window

Day -40

Email open rate drops from 38% to 12% for the at-risk cohort. No alert triggered.

Day -28

Site visit frequency halves. Session duration shortens by 60%. No action taken.

Day -14

Cart abandonment rate for this cohort hits 78%. Nobody flags the pattern.

Day 0

Last purchase. Customer churns. Win-back campaign begins 90 days later. Too late.

Signal Timeline vs. Intervention Window

Detection Gap
SignalDays Before ChurnDetection MethodAction Taken
Email decay40 daysAvailable in ESPNone
Visit drop28 daysAvailable in analyticsNone
Cart abandon surge14 daysAvailable in platformNone
Last purchase0 daysTracked as churnWin-back at day 90

Evidence Detected

Email decay ignored

Open rate dropped 26 points over 3 weeks before churn. No automated alert or intervention.

No behavioral scoring

Customer sessions shortening, carts being abandoned — no system aggregated these signals into a risk score.

Lagging metrics only

Team tracked purchase date as primary churn signal. By the time it moved, the decision was made.

Win-back too late

Recovery campaign launched at 90 days post-last-purchase. Optimal intervention window was 30-40 days.

Root Cause

The team tracked what customers did (purchased, didn't purchase) but not what they were signaling (engagement decay, behavioral drift). Every lagging indicator they watched told them the customer had already churned. The leading indicators they ignored told them it was about to happen.

"

Churn doesn't happen on the day of the last purchase. It happens on the day they stop opening your emails.

Case Conclusion

Invisible Churn Signal

Solved

Problem

Churn identified only after last purchase date exceeded 90 days. 40-day signal window missed.

Cause

Retention system tracked lagging indicators only. No leading signal monitoring or automated alerts.

Action

Behavioral risk score (email engagement + visit frequency + cart behavior). Automated intervention at score threshold.

Investigator's Takeaway

The brand was watching the wrong clock. Purchase date is what happened. Email engagement is what is happening. The churn signal was visible 40 days before anyone looked for it — and by then the customer was already gone.

The signals are already in your data. The question is whether your retention system is watching them.

Book a Retention Investigation →