Issue No. 17 | July 28, 2026

For warranty administrators who manage loss ratios, reinsurance, and contract performance.

This Week: What Blended Persistency Hides From Reserve Work

THE CURVE: 25%

Per Dark Sky Data’s cohort-based retention modeling, at an 8% annual cancellation rate, a book of contracts turns over roughly a quarter of itself every three years just to hold flat, before any new business growth is layered on. That compounding effect is easy to understate when persistency is reported as a single blended figure across the whole book.

Widen the lens to reserve adequacy and the same problem shows up. A blended persistency rate can sit at a stable-looking level while masking meaningfully different cancellation behavior across vintages, dealer channels, or term structures. Two books with identical headline persistency can carry very different cancellation patterns beneath the surface, leading to different reserve considerations.

THE ADMINISTRATIVE ANGLE

A single portfolio-level persistency number cannot tell you whether your book is stable or quietly deteriorating in specific cohorts.

Classifying contracts by cohort—new, retained, lapsed, and reinstated—rather than relying on an aggregate persistency figure reveals the customer behavior behind the headline metric. It applies the same diagnostic framework described in the source article: instead of asking only whether persistency appears stable, you can determine whether new originations are masking elevated cancellations in older vintages. Pairing those classifications with loss development and claims severity by the same cohorts answers that question directly.

A book with rising originations and flat blended persistency can still be masking a vintage with rising early-term cancellations, exactly the kind of shift that shows up in reserve adequacy before it appears in the headline loss ratio. The administrators better positioned for a reinsurance renewal are the ones who can already identify which vintage is driving the aggregate result rather than discovering it during the renewal process.

FROM THE BLOG

Customer Retention Metrics: Beyond the Headline Numbers

A blended retention figure describes the outcome of a book’s performance without explaining which cohorts produced it, whether that book contains warranty contracts, recurring service customers, or SaaS accounts. For administrators, the same cohort-level classification used to separate durable customers from masked churn applies directly to identifying stable vintages versus those showing cancellation patterns that may affect reserve adequacy.

THE RESERVE QUESTION

At 8% annual cancellation, a quarter of your book turns over every three years by default. If a reinsurer asked for your persistency broken out by vintage and term tomorrow, would the blended number be enough, or would the underlying cohorts tell a different story?

Until next Tuesday,

If you’d like to work through what cohort-level persistency and loss development data would show for your book, reply here and we’ll book 20 minutes.