Commentary on a 2026 ratemaking research call from the Casualty Actuarial Society, seeking proposals on the use of customer lifetime value in insurance pricing. Details are on the CAS website.
The CAS has invited researchers with customer segmentation expertise to propose work on using customer lifetime value in property-casualty insurance pricing. The call sits alongside a broader observation the CAS has been making: the role of the traditional pricing actuary is modernising to meet more diverse needs and more complex business problems, extending well beyond conventional loss cost and expense analysis.
Both statements are accurate. Together, they describe a shift that carries substantially more regulatory risk than the profession has collectively acknowledged — and it is worth being precise about where the risk actually sits, because the naive version of the concern is wrong.
What Customer Lifetime Value Means Here
Conventional ratemaking answers a single-period question: what premium covers the expected loss and expense of this policy, plus a provision for profit?
CLV asks a multi-period question: what is the expected present value of the entire relationship — this policy, its renewals, cross-sold products, referrals — net of acquisition and servicing costs?
The two answers differ, and the components of the difference are worth separating because they are not equally contentious.
Retention probability. A customer likely to renew for a decade generates value across ten policy periods. Acquisition cost amortises over ten years rather than one.
Cross-sell propensity. A motor customer likely to add home and umbrella coverage is worth more than an identical motor customer who will not.
Loss ratio trajectory. Loss experience is not stationary across tenure. New business frequently runs worse than renewal business at the same rating profile, for reasons including selection effects and rating error at inception.
Servicing cost trajectory. Claims frequency, contact volume and channel preference vary systematically with tenure.
Price elasticity. How the customer's renewal probability responds to a rate change.
Note the last one. That is where the whole thing turns.
The Distinction That Determines Whether This Is Acceptable
There is a real difference between two things that get lumped together, and getting it right matters enormously.
Cost-based multi-period pricing recognises that the cost of serving a customer differs across the relationship. Amortising acquisition cost over expected tenure, reflecting the observed loss ratio differential between new and renewal business, accounting for lower servicing cost among long-tenure digital customers. These are cost recognitions. They happen to require a multi-period view, but they are conventional actuarial reasoning applied over a longer horizon.
Demand-based price optimisation sets price according to what the customer will tolerate rather than what they cost. Charging more to a customer whose modelled elasticity is low — because they do not shop, or find switching difficult — regardless of their expected cost.
The first is defensible and, in my view, actuarially correct. The second has been the subject of significant regulatory action.
The UK is the clearest case. The FCA's general insurance pricing practices rules addressed the "price walk," under which renewing customers were charged progressively more than equivalent new customers, and now require that a renewing customer's price be no higher than the equivalent new business price for the same customer through the same channel. That is a direct regulatory prohibition on the most common form of elasticity-based pricing.
In the US, several state regulators have restricted price optimisation, and the NAIC has addressed the practice, though treatment varies materially by state.
CLV pricing that reflects cost differences is broadly acceptable. CLV pricing that reflects willingness-to-pay is, in significant jurisdictions, not. Any CLV research or implementation that does not draw this line explicitly and early will produce something unimplementable in regulated personal lines.
Where the Fairness Problems Hide
Even the defensible cost-based version carries fairness exposure, in ways less obvious than the elasticity issue.
Retention correlates with characteristics you cannot price on. Customers who shop around are systematically different from those who do not — by age, digital access, financial pressure, language, and time availability. A model that rewards long expected tenure with a lower price may be rewarding the absence of alternatives. Where switching costs fall disproportionately on disadvantaged groups, a tenure-based discount is a disadvantage-based discount wearing a respectable coat.
Cross-sell propensity correlates with wealth. The customer likely to buy umbrella and valuables coverage has assets to protect. A CLV model that prices motor insurance more favourably for customers with high cross-sell propensity is offering lower motor premiums to wealthier customers. Whether that is objectionable is a policy question; that it will be noticed is not in doubt.
Multi-period models are far harder to explain. Adverse action and consumer explanation requirements assume a decision traceable to identifiable factors. Explaining a price that reflects a ten-year modelled relationship — including a retention model, a cross-sell model and a cost trajectory — to a customer or a regulator is materially harder than explaining a conventional rate. Under the UK Consumer Duty, firms must be able to articulate in terms accessible to a customer why a decision was reached. A CLV price is not naturally articulable in those terms.
Building It Defensibly
Six design principles.
Separate cost components from demand components architecturally. Not as a documentation convention — as a structural property of the system, so that the demand model can be switched off entirely without breaking the cost model. This is the single most valuable design decision, because it means a regulatory challenge to elasticity pricing does not invalidate your whole pricing structure.
Cap the CLV adjustment. Whatever the model says, bound the effect. An unbounded multi-period adjustment will eventually produce an indefensible individual outcome, and one indefensible outcome in a complaint file is worth more to a regulator than a thousand reasonable ones.
Test for demographic gradient at the CLV component level, not just the final price. The final price may look acceptable in aggregate while the retention component carries a strong socioeconomic gradient that the loss cost component happens to offset. Component-level testing catches this; end-to-end testing does not.
Validate the retention model against actual behaviour, repeatedly. Retention models degrade faster than loss models because consumer behaviour changes faster than physical risk. A retention model calibrated pre-2022 has almost certainly not survived the inflation environment.
Document the fairness definition and the elasticity position explicitly. State in the pricing methodology whether demand elasticity is used, and if so how it is bounded. Having this written down before a supervisor asks is worth a great deal.
Run the availability analysis. Which customer segments does CLV pricing push out of the market? That question will be asked eventually, and having the answer already is a much better position than commissioning the analysis in response to a challenge.
On the Modernising Pricing Actuary
The CAS observation about the expanding pricing role is right, and I would add a caution.
As pricing actuaries take on retention modelling, marketing analytics, customer segmentation and lifetime value, they move into territory where the professional framework is thinner. Actuarial standards on ratemaking are well developed. Standards on demand modelling and customer segmentation are not — because until recently those were marketing functions, not actuarial ones.
That gap creates an opportunity and a hazard. The opportunity is that actuaries bring documentation discipline, validation rigour and professional obligation to work that previously had none of these. The hazard is that actuaries acquire professional responsibility for outputs the profession has not yet developed standards to govern.
Both point the same way: the expanded role is a good development, provided the profession extends its standards to cover it rather than allowing the standards boundary to sit where the job description used to.
Key Takeaways
- The CAS is seeking research on customer lifetime value in P&C pricing, alongside a broader observation that the pricing actuary's role is expanding beyond conventional loss cost analysis.
- Cost-based multi-period pricing — amortised acquisition cost, tenure-varying loss and servicing costs — is defensible; demand-based elasticity pricing is restricted in the UK and several US states.
- Separate cost and demand components architecturally so the demand model can be disabled without breaking the pricing structure.
- Retention correlates with switching ability and cross-sell propensity correlates with wealth, so CLV components carry demographic gradients even when the final price looks neutral.
- Multi-period pricing is substantially harder to explain to consumers, creating tension with Consumer Duty-style articulability requirements.
Frequently Asked Questions
What is customer lifetime value pricing in insurance? Pricing that reflects the expected present value of the whole customer relationship — renewals, cross-sold products, tenure-varying loss and servicing costs, acquisition cost amortisation — rather than the expected cost of a single policy period. It differs from conventional ratemaking primarily in horizon.
Is customer lifetime value pricing legal? It depends on the components used and the jurisdiction. Reflecting genuine cost differences across a customer relationship is generally acceptable. Pricing based on modelled willingness to pay is restricted — the FCA's general insurance pricing practices rules address renewal price walking directly, and several US states restrict price optimisation.
How does CLV pricing interact with the Consumer Duty? The Consumer Duty requires firms to be able to explain decisions in terms a customer can understand, and to demonstrate fair value. A price reflecting a modelled multi-year relationship is inherently harder to articulate than a conventional rate, so firms using CLV components need a customer-facing explanation designed alongside the model rather than retrofitted.
Jonas Osman Abdelghafour is the founder of Quantica Risk, which builds pricing, retention and fairness-assessment models for insurers in the UK and internationally. This article is commentary on a publicly available CAS research call and is not legal advice.