Overview of the specialty
Actuarial marketing applies the actuary's tools (present value, persistency probabilities, risk and behavioral models) to the commercial decisions of an insurer or an intermediary: whom to make an offer to, at what price, through which channel, with what incentive and how much to invest in retention. Its unit of analysis is not the one-year policy but the entire customer relationship, including renewals, cross-selling, acquisition cost and expected claims.
It exists because sales and technical teams tend to measure different things. Marketing celebrates written premium and conversion rates; the technical side looks at loss ratios and rate adequacy. A campaign that doubles sales by attracting bad risks, or a discount that retains customers who would have stayed anyway, destroys value even if its commercial metrics look good. The actuary translates both views into a single metric: the economic value each decision creates.
The technical core is demand. The expected cost model (what claims the customer will generate) is combined with a response model (the probability that the customer buys or renews at each price). With both, one can estimate elasticity, optimize renewal prices, decide whom to target with a retention offer and measure the real effect of campaigns with controlled experiments.
Mexico has its own particularities. Bancassurance and mass-market insurance sold by retailers, finance companies and digital platforms carry great weight in life, accident, property and credit-related lines, and are governed by specific rules on distribution through legal entities. The LISF prevents discounts from reducing the net risk premium, the financial transparency law prohibits tied selling and CONDUSEF (the financial consumer protection agency) supervises advertising and sales practices. Price optimization here is always constrained optimization.
Key functions
Customer lifetime value (CLV)
Calculates the present value of the future margins a customer will generate, taking into account renewal probability, expected claims, expenses, commissions and cross-selling. CLV determines how much to pay to acquire a customer in each channel and how much to invest in keeping them.
Elasticity and price optimization
Estimates how the probability of purchase or renewal changes with price, by segment, and searches for the prices that maximize portfolio value subject to constraints: technical adequacy, caps on increases, fairness rules and volume targets. Documents the separation between the cost of risk and the commercial component.
Customer segmentation
Groups customers by value, risk and behavior using supervised methods (trees, value models) and unsupervised ones (k-means, Gaussian mixtures). Good segmentation distinguishes the profitable, loyal customer from the profitable but price-sensitive one, who call for opposite strategies.
Retention and uplift models
Distinguishes between the probability of lapsing and the incremental effect of an action on that probability. Uplift models identify persuadable customers, whose decision changes with the offer, and avoid spending incentives on those who would stay anyway or will leave no matter what.
Cross-selling and channel profitability
Estimates the propensity to buy a second product (life insurance for an auto customer, medical expense insurance for a bank customer) and the profitability of each channel: agents, brokers, bancassurance, retailers, direct and digital sales. Brings together commissions, acquisition cost, persistency and claims by channel, because the risk mix differs depending on who sells.
Experimentation and attribution
Designs A/B tests with a large enough sample to measure the effect of a price, an offer or a message, and attribution models that share credit for a sale among prior touchpoints. Without a control group, almost any campaign looks successful.
Techniques, models and formulas
The techniques combine financial mathematics, survival models, demand econometrics and experimental design. The golden rule is to keep the two models separate: the expected cost model, which belongs in the technical note and must be adequate, and the demand model, which guides commercial decisions within the margins the first one allows.
CLV models
From the closed-form formula with constant margin and retention to cohort-level and customer-level models that project renewals, claims and additional purchases. In insurance, the margin must net out the customer's expected claims; premium alone does not measure value.
Demand and elasticity models
Logistic regressions for conversion and renewal with the price change as an explanatory variable, controlling for risk, channel and competition. The slope with respect to price gives the elasticity by segment; the quality of the estimate depends on there being price variation not explained by risk.
Constrained optimization
Nonlinear programming that maximizes the expected value of the portfolio subject to caps on increases, a technical floor, fairness rules and retention targets. In practice it is solved by segment using price grids or gradient methods.
Uplift models
They estimate the causal effect of an action from experimental data, using two models (treatment and control), interaction models or causal forests. They are evaluated with Qini curves rather than AUC, because what matters is the incremental effect.
A/B testing and experimental design
Random assignment of customers to different prices or offers, with a sample size calculated to detect the minimum relevant effect. Detecting an increase in retention from 65% to 70% at 5% significance and 80% power requires about 1,374 customers per group.
Channel attribution
Last-touch, position-based or data-driven models (Markov chains, Shapley values) that allocate credit for a sale among touchpoints. They make it possible to set channel investment based on actual contribution rather than the last interaction.
m is the annual margin per customer (premium minus claims, expenses and commissions), r the annual renewal probability and d the discount rate. The margin is collected at the start of each year, including the current one.
Q is the number of policies sold or renewed and P the price. With ε = -1.5, a 2% price increase reduces the quantity by about 3%.
qi(Pi) is the probability of purchase or renewal at price Pi, Ci the expected cost (claims and expenses), P̲i the technical floor that respects the net risk premium, Pi0 the current price and δ the cap on price changes. Fairness constraints between new and renewing customers can be added.
Regulatory framework in Mexico
LISF art. 200 · The net risk premium is not negotiable
Net risk premiums are set on a technical basis, and bonuses or dividends cannot reduce them. Every commercial discount, promotion or elasticity-based adjustment must be funded from the expense and profit loadings: the floor of any optimization is the risk premium in the technical note.
LISF arts. 102 and 103 · Distribution through legal entities
For insurance written under standard-form contracts (except social security pensions and surety bonds), the policy can be taken out through a legal entity without an agent, such as a bank or a retailer. The insurer may pay it for services under contracts registered with the CNSF (National Insurance and Surety Commission), and those entities are subject to its inspection and oversight. This is the legal basis for bancassurance and mass-market insurance.
LTOSF art. 23 Bis · Ban on tied selling
The Ley para la Transparencia y Ordenamiento de los Servicios Financieros (Financial Services Transparency Law) prohibits making a financial transaction or service conditional on taking out another. In addition, for personal and revolving loans the same law requires (art. 18 Bis 5) that, when an optional insurance product is offered at the customer's expense, its cost, the fact that it is optional and the fact that cancelling it does not cancel the loan must be stated expressly. Cross-selling strategies must respect this boundary.
CONDUSEF · Sound practices, transparency and advertising
CONDUSEF's rules on sound practices, transparency and advertising for insurance companies govern product information and advertising. The Public Registry of Users (REUS) lets people block promotional calls and messages from financial institutions, which limits campaign contact lists.
LFPDPPP 2025 · Data for marketing
Using personal data to profile, segment and contact customers requires a privacy notice with specific purposes and respect for the right to object, particularly with respect to automated decisions that significantly affect the data subject.
International benchmarks on pricing fairness
Since January 1, 2022, the UK FCA has prohibited charging renewing motor and home insurance customers more than an equivalent new customer would pay. In 2015 the NAIC examined price optimization in a white paper, and several US states restricted the use of non-risk factors. They are useful benchmarks for setting internal fairness policies.
Case study: across-the-board discount or targeted retention with uplift?
An auto insurer has a segment of 5,000 policies with a high probability of lapse. Average premium: $9,500. Annual margin per policy after claims, expenses and commissions: 12%, that is, $1,140. Expected retention without action: 65%. Discount rate: 10%. The net risk premium is 62% of the premium ($5,890). Marketing proposes a permanent 4% discount ($380) for the whole segment, which it estimates will raise retention to 75%. The figures are illustrative.
- CLV without action: $1,140 × 1.10 / (1.10 - 0.65) = $1,140 × 2.4444 = $2,786.67 per policy.
- CLV with the across-the-board discount: the margin falls to $1,140 - $380 = $760 and retention rises to 75%. CLV = $760 × 1.10 / (1.10 - 0.75) = $2,388.57. Each policy loses $398.10 in value; across the 5,000 policies, the campaign destroys $1,990,476 even though retention improves by ten points.
- Alternative: a prior A/B test (about 1,374 customers per group to detect a change from 65% to 70%) feeds an uplift model, which identifies 1,500 persuadable customers whose retention goes from 60% without the offer to 80% with it.
- CLV of the persuadables: without the offer, $1,140 × 1.10 / 0.50 = $2,508.00; with the offer, $760 × 1.10 / 0.30 = $2,786.67. The gain is $278.67 per customer, $418,000 in total. After deducting the contact cost of $150 per customer ($225,000), the net value created is $193,000.
- Regulatory and fairness check: the discounted price ($9,120) remains well above the net risk premium ($5,890), so the discount is funded from loadings as the LISF requires. The criterion for allocating the offer is documented to show that it does not use prohibited variables or penalize the loyalty of customers who would not lapse.
Where they work and how they grow
Builds customer databases, measures conversion, retention and profitability by channel and campaign, and calculates CLV by segment. Learns to reconcile commercial figures with technical and accounting ones.
Estimates demand and elasticity models, designs A/B tests and uplift models, and proposes renewal price adjustments and targeted offers along with their expected effect on value.
Coordinates pricing and retention strategy for a line of business, negotiates the limits of optimization with sales and technical teams, and is accountable for the value created by campaigns and channels.
Defines the strategy for profitable growth, the channel mix and bancassurance partnerships, and pricing governance with fairness and compliance criteria before the CNSF and CONDUSEF.
- Insurers with commercial pricing, CRM or customer intelligence teams
- Banks and financial groups with bancassurance operations
- Agents and brokers with portfolio management and customer analytics
- Digital platforms, comparison sites and insurance aggregators
- Retailers, finance companies and telecom companies that sell mass-market insurance
- Strategy, pricing and analytics consulting firms
- Afores (retirement fund managers), in retention and account transfers
Skills and tools
Technical
Tools
Personal
Certification and training
CONAC certification in Life and Health or Property and Casualty
Provides the technical standing to sign the technical note that sets the floor for any commercial optimization. See Certification.
CAS Institute · CSPA
The predictive analytics credential from the CAS affiliate covers modeling and an applied project, both useful for conversion, retention and demand models.
SOA · PA and ATPA exams
The predictive analytics exams on the path to ASA strengthen the foundation for response and retention models.
Graduate programs in marketing analytics or data science
Master's degrees in data science, applied economics or quantitative marketing provide causal inference, experimental design and demand econometrics.
Trends reshaping the field
Fairness in renewal pricing
The British ban on charging renewing customers more than new ones has become a global benchmark. Mexican insurers are starting to set internal policies that limit the gap between renewal and new business prices, to get ahead of conduct supervision.
Embedded insurance and digital partnerships
Insurance is offered as part of buying a car, a plane ticket, a phone or a digital loan. The actuary measures the profitability of each partnership, the channel's loss experience and the value of the customers who arrive that way.
Causal inference instead of correlation
Commercial teams are moving from propensity models to uplift models and continuous experimentation, because the business question is what happens if I act, not who is most likely to buy.
Personalization with limits
Models allow almost individual offers, but personal data regulation, conduct rules and reputation impose limits. Designing pricing governance, with documented criteria and fairness reviews, is part of the actuarial job.
Frequently asked questions
What does an actuary do in marketing if not design campaigns?
Is price optimization legal in Mexico?
Why is the CLV of an insurance customer different from that of a retail customer?
How large should an A/B test be?
Further reading
- Ley de Instituciones de Seguros y de Fianzas, arts. 102, 103 and 200 (Chamber of Deputies)
- Ley para la Transparencia y Ordenamiento de los Servicios Financieros (Chamber of Deputies)
- CONDUSEF, rules on sound practices, transparency and advertising applicable to insurance companies (DOF 2015)
- CONDUSEF, Public Registry of Users (REUS)
- FCA, measures against the loyalty penalty in home and motor insurance
- NAIC CASTF, Price Optimization White Paper (2015)
- Ley Federal de Protección de Datos Personales en Posesión de los Particulares (Federal Law on the Protection of Personal Data Held by Private Parties)