How to measure protection clawback by cohort

The problem with the monthly bill

The clawback deducted in any month relates to policies sold across many earlier months, each at a different point in its earnings period. A rise or fall in the monthly figure might reflect a change in sales volumes a year ago, a single large lapse or a genuine change in persistency. Without separating these, it is very hard to tell whether the business is improving.

Many firms track clawback only as a line in the management accounts. That is useful for cash flow, but it gives little help in deciding what to change.

What a cohort is

A cohort is a group of policies sold in the same period, typically a calendar month. Measuring each cohort separately shows how many of its policies were still in force after one, three, six, twelve, twenty-four and forty-eight months, and how much commission was reclaimed from it. Comparing cohorts at the same age is like-for-like, so differences reflect real changes rather than timing.

Building the view

You need a record of each policy with its sale date, product, premium, commission paid, adviser, lead source and insurer, plus the date and reason for any cancellation and the amount reclaimed. Most distributors have this data across their CRM and commission statements. The analysis can start in a spreadsheet: one row per cohort, one column per month since sale, showing the share of policies still in force.

What to look for

First, the shape of the curve. Steep early drops point to problems at or just after the sale. Gradual later declines point to affordability and disengagement. Second, differences between groups. Break cohorts down by product, adviser, lead source and insurer. A lead source with consistently worse persistency may be costing more in clawback than it earns. An adviser whose customers lapse early may need support with suitability or explanation.

Reasons, not just rates

Where reasons for cancellation are available, add them. Separate lapses the firm could reasonably have influenced, such as failed payments not followed up, documents never received or replacement errors, from those it could not, such as a mortgage being repaid. The influenceable share is the realistic target for improvement.

Measuring the effect of a change

When you introduce something new, such as a post-sale check, compare the cohorts before and after at the same ages. Within a cohort, you can also compare customers who completed the check with those who did not. Across anonymised Adviser.ai client data, customers who complete a post-sale check have been around 50% less likely to cancel. A cohort view lets you see whether that holds in your own book, and how quickly the benefit appears.

Forecasting

Once you have several years of cohort curves, you can forecast future clawback from current sales. That helps with cash flow planning and makes the commercial value of retention activity clear: if a change improves the twelve-month in-force rate by a few points, the forecast shows the commission that will no longer be reclaimed.

Sharing the results

Cohort data is most useful when people see it. Share the curves with advisers, sales managers and service teams, and show how their actions affect them. Report the headline measures to the board alongside other Consumer Duty outcome data. When everyone can see how persistency is moving, retention becomes part of how the business runs rather than a back-office concern.

Common pitfalls

Mixing policies with different earnings periods in one curve, which distorts comparisons. Comparing cohorts at different ages. Ignoring small groups whose results swing widely from month to month. Treating every cancellation as equal when some were reasonable. And producing the analysis once rather than regularly. A simple cohort view updated monthly is far more useful than a detailed one produced once a year.

Questions the cohort view answers

Is persistency improving or getting worse? Which products, advisers, lead sources and insurers carry the most clawback risk? How much of our clawback comes from the first three months? Did last quarter's process change make a difference? How much clawback should we expect next year? Each answer points to a specific action.