Customer retention analytics

Customer retention analytics for service businesses usually starts in the wrong place: marketing dashboards. If invoices are your system of truth, retention should be read from purchase history—who still buys, who is slowing, and who has gone quiet relative to their own pattern.

What to measure from the books

• Repeat rate. Share of customers with more than one invoice in a lookback window.
• Cadence. Typical gap between invoices per customer.
• Recency vs cadence. Time since last invoice compared with that customer’s own gap.
• Value at risk. Prior-year spend for accounts now overdue on their rhythm—not a blunt “90 days silent” rule.

A simple retention reading

1. Export invoices with contact and dates (and lines if you can).
2. Per contact, compute invoice count and median gap.
3. Flag contacts whose silence exceeds ~1.5× their median gap.
4. Separate true one-offs from repeaters before you call anything “churn.”

Honest limits

Invoices will not tell you why someone left, whether a site still needs service, or if a competitor won the work—unless that is written into evidence you trust. Keep unknowns labelled; do not invent retention scores from thin data.

How Spearmint helps

Spearmint surfaces quiet and slowing accounts on a Revenue Map with source references, and Mint can explore cohorts such as customers buying less frequently. You still own the relationship work; Spearmint finds who deserves attention.