Persistency Analysis: How Insurers Measure Policy Retention

In short: Persistency analysis tracks the proportion of insurance policies that remain in force over time. Lapse rates are highest in the first two policy years and decline to a steady state from year three. The 13-month and 25-month ratios are the industry standard KPIs, driven by commission clawback structures that align distributor incentives with customer outcomes. A cohort table built in Excel applies assumed lapse rates recursively to track in-force counts and cumulative persistency at every duration.

Every insurer sells a policy on the basis of a long-term relationship: the policyholder pays a regular premium, and in return the insurer pays a claim if the agreed event occurs — death, disability, critical illness, or income loss. The insurer’s profitability depends entirely on that premium stream continuing long enough to recover what it cost to acquire the business in the first place.

The problem is that policyholders do not always keep paying. Life changes. A job is lost. Premiums feel unaffordable against a rising cost of living. A competitor offers cheaper cover. A direct debit fails after a house move and no one notices until the grace period has passed. When a policy stops paying premiums and has no accumulated surrender value, it lapses. The insurer has paid commission to the adviser, absorbed underwriting and administration costs, and set up the policy on its systems — and now receives nothing further from that customer.

Persistency analysis is the systematic process of measuring, monitoring, and understanding lapse behavior across a book of business. It sits at the intersection of finance, actuarial science, and regulatory conduct: the numbers feed directly into pricing assumptions, embedded value calculations, and the FCA’s Consumer Duty assessments. This post covers the mechanics from first principles.

The Lapse Rate and the Persistency Rate

Two numbers describe the same reality from different angles.

The lapse rate for a given policy year measures the proportion of in-force policies that exit during that year:

Lapse Rate (Year n) = Policies that lapsed during Year n
                      ─────────────────────────────────────
                      Policies in force at the start of Year n

The persistency rate at a given duration measures the cumulative proportion of originally sold policies that are still in force at that point:

Persistency Rate (Month 13) = Policies still in force at Month 13
                               ─────────────────────────────────────
                               Policies sold at inception

These two numbers are linked. If year-1 lapse rate is 10% and year-2 lapse rate is 6%, the 25-month persistency rate is approximately:

(1 − 0.10) × (1 − 0.06) = 0.90 × 0.94 = 84.6%

For a starting cohort of 1,000 policies with lapse rates of 10%, 6%, 4%, and 3% in years one through four:

In Excel, the in-force count for each period is a simple recursion:

In-force (end of Year n) = In-force (start of Year n) × (1 − Lapse Rate n)
Cumulative Persistency % = In-force (end of Year n) / Original Cohort Size

The characteristic shape of this curve — steep decline in year one, flattening from year three — is consistent across almost every insurance product and market. New policyholders who are going to lapse typically do so quickly.

The Industry’s Standard KPIs: 13-Month and 25-Month Persistency

The insurance industry reports persistency at two specific durations above all others, and the choice is not arbitrary — it is tied directly to how distribution commission is structured.

When an independent financial adviser (IFA) or protection specialist sells a life or protection policy, they receive an initial commission payment from the insurer, typically at or shortly after policy inception. For a protection product, this commission can equal 100% or more of the first year’s annual premium. The insurer recoups this outlay over years of future premium income at the margin between premium received and the cost of providing cover.

To prevent distributors from prioritizing commission income over customer suitability, UK regulation requires commission clawback: if a policy lapses within a defined period — usually 12 months — the distributor must return all or a pro-rata portion of the initial commission to the insurer. A policy that reaches its 13th monthly payment has therefore survived one full year plus one month, confirming that:

1. The clawback obligation has expired — the adviser has earned the commission.
2. The insurer has received 13 months of premium income and begun recovering acquisition costs.
3. There is at least basic evidence that the policy met the customer’s needs — they kept paying for a year.

The 25-month persistency rate marks the end of a second clawback window operated by some insurers, or simply the end of the second policy anniversary — by which point the insurer will typically have recovered the majority of its acquisition costs on surviving business.

UK market benchmarks for protection products (term assurance, critical illness, income protection):

A 13-month persistency rate below 80% is a regulatory concern. Under the FCA’s Consumer Duty framework, insurers must demonstrate that their products deliver fair value to customers throughout the policy term — sustained high lapse rates are evidence to the contrary and can trigger supervisory engagement or product reviews.

Why Policies Lapse

Lapse behavior is not random noise. It clusters around specific, identifiable triggers. Understanding those triggers is what converts raw persistency statistics into actionable management information.

Affordability shocks. The most common driver across all product types. A policyholder who loses their job, faces a rent increase, or encounters any significant reduction in disposable income will review their outgoings. Insurance premiums — particularly protection products where the benefit feels intangible — are often canceled before gym memberships. The UK cost-of-living pressure between 2022 and 2024 pushed year-1 lapse rates 2–4 percentage points higher on many protection books.

Mis-selling and poor needs assessment. Policies placed with customers who did not fully understand what they were buying, or whose circumstances have changed significantly since the sale, lapse at materially higher rates. A customer who was sold critical illness cover they cannot comfortably afford, or income protection that duplicates a benefit already provided by their employer, is likely to cancel as soon as they think carefully about the policy. This category is of acute regulatory concern — the FCA’s supervisory focus on protection markets has included explicit analysis of whether high-lapse distributors correlate with poor customer outcomes.

Competitor replacement. A customer receives a re-quote from a competing insurer — often driven by price comparison sites or a new adviser — and finds cheaper cover. They cancel the existing policy and start a new one. This is most common in term assurance, which is a commodity-like product easily compared across providers, and least common in products with built-up benefit entitlements or health history disclosures that make switching costly.

Administrative lapse. A direct debit fails because a bank account was closed, an account number was not updated after switching banks, or the policyholder moved house and forgot to notify the insurer. Administrative lapses are economically distinct from genuine cancellations — a meaningful proportion can be recovered through proactive reinstatement outreach within the first 60 days. Insurers with strong reinstatement processes report 20–35% recovery rates on administrative lapses, which materially improves effective persistency.

Life events. A divorce dissolves a joint life policy. A house sale prompts a policyholder to cancel a decreasing term policy tied to a mortgage that no longer exists. A bereavement claim on one policy leads the surviving partner to reassess their own cover. These events are unpredictable at the individual level but statistically stable in aggregate, and pricing teams model them as a component of the long-term lapse basis.

Building the Persistency Cohort Model in Excel

A cohort persistency model takes a group of policies written in a defined period, applies assumed annual lapse rates by channel and policy year, and tracks in-force counts and cumulative persistency at each duration. Here is a complete structure for 5,000 policies sold in Q1 2025.

Tab structure:

Assumptions tab — lapse rates by policy year and channel:

IFA-distributed business persistences materially better than direct-sold. The ongoing adviser relationship reinforces the value of cover and catches administrative lapses before they become permanent — a customer whose direct debit fails is more likely to be contacted by their adviser than by an insurer’s call center. Direct-channel lapse rates in year one can run 50–70% higher than IFA rates on comparable products.

Cohort Model tab — annual in-force table for the 5,000-policy cohort (split: 60% IFA, 30% direct, 10% workplace):

The in-force recursion for each channel in Excel (using named ranges):

=IFA_Year0 * (1 - IFA_Lapse_Y1)           ← Year 1 end IFA count
=IFA_Year1 * (1 - IFA_Lapse_Y2)           ← Year 2 end IFA count
...
Persistency % = Total_YearN / Total_Year0

The blended 13-month persistency (proxied by year-1 end) of 89.8% sits within market norms. The weakness is the direct channel — 86.0% — which is below the market average for a product of this type. The workplace channel at 94.0% is exceptionally strong, driven by employer endorsement and payroll-linked premiums that remove the direct-debit failure risk entirely.

Interpreting Results: What Good Looks Like

Raw persistency numbers only carry meaning relative to a benchmark. Use three reference points simultaneously:

Internal cohort trend. Compare each new cohort’s 13-month persistency against its predecessor. If the IFA channel persistency has declined from 93% to 90% to 87% over three cohorts, that is a structural deterioration — not noise. The cause is most likely in the distributor mix (new distributors added who have a different quality of business) or a product change that has created a suitability mismatch.

Channel decomposition. A blended 90% persistency rate can conceal a 94% IFA rate and an 82% direct rate — two entirely different stories with different management actions. Never present or discuss the blended rate without the channel split. The most useful display format is a side-by-side bar chart by channel at 13 months and 25 months for each annual cohort.

Industry benchmark. The Association of British Insurers (ABI) publishes aggregate market persistency data annually. The FCA incorporates persistency benchmarking into its Consumer Duty supervisory assessments — firms with 13-month persistency more than five percentage points below the market average for comparable products are at elevated risk of supervisory scrutiny.

A persistency heat map is the standard board-level reporting format. Rows are policy cohorts (e.g. Q1 2023, Q2 2023, …). Columns are duration (13 months, 25 months, 37 months, …). Cell values are persistency rates. Conditional formatting highlights deteriorating cells amber or red. This makes it immediately visible whether a deterioration is cohort-specific (one quarter’s sales were poor quality) or duration-specific (the whole book is lapsing at a particular anniversary).

Persistency and New Business Profitability

The reason persistency occupies so much management attention is that the insurer’s profit model depends on it completely. New business is written at a loss in year one: acquisition costs — commission, underwriting, new business administration — typically exceed the first year’s premium for a standard protection policy. The insurer recovers those costs through the margin between premium received and the cost of claims provision and ongoing administration in future years.

A simplified break-even illustration for a term assurance policy at £30 per month:

Break-even occurs at the end of year two. Any policy that lapses before that point is loss-making in aggregate. At 89.8% year-1 persistency on a cohort of 5,000 policies:

• 510 policies lapse in year 1 — each costs the insurer £210 in unrecovered acquisition costs: £107,100 total
• 285 policies lapse in year 2 — each breaks even but generates no profit
• The remaining 4,205+ policies bear the entire profitability of the cohort

If year-1 persistency dropped from 89.8% to 84.0% — a 5.8 percentage point deterioration — an additional 290 year-1 lapses would cost another £60,900 on this cohort alone. Across a book of 100,000 new policies per year, a 3-point persistency deterioration translates to several million pounds of embedded value destruction annually.

This arithmetic explains why persistency is treated as a leading indicator of insurer financial health — it is visible months or years before the consequences show up in reported profit.

Key Takeaways

• Persistency measures what proportion of policyholders are still paying premiums at a given duration. The annual lapse rate is its complement.
• 13-month and 25-month persistency are the standard KPIs because they align with commission clawback windows — the mechanism the FCA uses to align distributor incentives with customer outcomes.
• Lapse rates follow a predictable shape: high in years one and two, declining to a lower steady state from year three or four. Direct-sold business consistently lapses faster than IFA-intermediated business.
• Build the cohort model in Excel with separate columns per distribution channel and apply lapse rates recursively to track in-force counts and cumulative persistency.
• Always decompose the blended persistency rate by channel before drawing any conclusions — the blended number hides the most actionable information.
• Persistency is a leading profit indicator: a policy that lapses before break-even duration destroys value. A 3-point deterioration in year-1 lapse rates can cost a mid-sized insurer several million pounds in embedded value on a single year’s cohort.

Practice

Build a 10-year annual cohort model for a life insurer that wrote 2,000 policies in Q1 2025. Channel split: 60% IFA (year 1–5+ lapse rates: 8%, 5%, 3%, 2.5%, 2%), 30% direct (15%, 10%, 6%, 4%, 3.5%), 10% workplace (5%, 3%, 2.5%, 2%, 2%). Calculate the 13-month and 25-month persistency rates for each channel and in aggregate. Then run a stress scenario where direct-channel year-1 lapse rate rises to 22% due to cost-of-living pressure. Quantify: (1) the additional year-1 lapses versus base case; (2) the lost cumulative premium income over five years assuming a £35 monthly premium per policy; and (3) the unrecovered acquisition cost assuming £210 per year-1 lapse.