The situation

Harbor Group managed commercial property and liability policies for mid-market businesses. Account managers handled between 40 and 80 accounts each, with renewal conversations typically beginning 60 days before policy expiration. The retention rate had declined from 82% to 71% over three years, costing roughly $3.4M in annual recurring revenue.

The pattern that frustrated leadership most: the churned accounts hadn't expressed dissatisfaction. No formal complaints, no escalations, no service failures the team could point to. They simply received notice of non-renewal, often from a broker they hadn't known was in conversation with their client.

What we found

We extracted account activity data, email engagement logs, and claim history for all accounts over the previous three years and built a cohort analysis: accounts that renewed versus accounts that churned, looking backward from the renewal date for distinguishing signals.

Three patterns appeared consistently in churned accounts, all visible 90 or more days before cancellation. First: the primary contact had changed (new CFO, new risk manager, or new procurement lead) without the account manager updating their contact record or initiating a relationship-building conversation. 64% of churned accounts had a contact change in the preceding six months. Only 12% of retained accounts had the same pattern.

Second: claim frequency had dropped to zero in the 12 months before renewal. Counterintuitively, accounts with no claims were more likely to churn than accounts with moderate claims — because zero-claim accounts had no recent touchpoint where value was demonstrated. The account manager's work had become invisible.

Third: email engagement from the account manager to the client had dropped below one meaningful touchpoint per month in the 90 days before renewal in 71% of churned accounts. The relationship had gone quiet at exactly the window when competitors were most likely to be making contact.

What changed

A health score model was built incorporating all three signals and implemented in the CRM as a renewal risk dashboard. Account managers saw their at-risk accounts 90 days before renewal with the specific factor driving risk for each one.

Contact change alerts were automated: any update to the primary contact field triggered a task to schedule a relationship call within 14 days. Zero-claim accounts triggered a quarterly value review call to ensure visibility was maintained. Low-engagement accounts triggered an outreach sequence starting 90 days before renewal.

Churn dropped from 29% to 19% in the first year, a 34% reduction. Renewal retention moved from 71% to 86%. The most significant change was behavioral: account managers stopped treating renewals as administrative events and started treating the 90-day window as an active retention period. The data had made the stakes of that window visible in a way that anecdotal feedback never had.