Products inherit the incentives of those who make them.

Insurance reaches a household through a chain. An insurer carries the risk and sets the price. A distributor, such as a bank, a broker or an agent, sells the policy and is usually paid by commission. A comparison site often sits in front of both and earns from the traffic it gathers.

Each link acts rationally. The insurer builds cover that keeps its risk and its margins on target. The distributor favors the products that pay best for the effort a sale takes. The comparison site ranks offers in ways that sustain its traffic. None of this requires bad faith. It is what each business model rewards.

Much of French insurance moves through intermediaries. In 2023, brokers distributed 23.6% of non-life premiums and banks 18.5%³.

The person who buys the policy stands at the end of the chain. They must compare guarantees and read exclusions that were defined upstream, for other objectives. They are the customer. The product was often optimized for someone else.

The clearest public measurement of this mechanism concerns credit protection insurance sold by banks alongside loans. On 30 September 2022, EIOPA, the European insurance supervisor, published a warning on the subject, based on data from 174 insurers and 145 banks covering 2018 to 2020¹.

Another finding completes the picture. 74% of the banks had no cost allocation model, so their commissions could not be shown to reflect the cost of the service.

These figures describe a product shaped around the party that sells it. When the distributor is paid a share of the premium, the product that earns a place on the shelf is the one that pays the distributor well. It is not necessarily the one that pays claims well.

Since 2022, French law has allowed borrowers to replace the insurance on a home loan at any time⁴. Borrowers used the right. Requests to switch rose from 198,530 in 2021 to 496,654 in 2024².

The market moved much less. Insurers outside the bank’s own channel covered 16% of home loans in 2021 and 17.48% in 2024². Fewer than one home loan in five is insured outside that channel. For scale, borrowers paid €6.83 billion in 2024 for death, disability and incapacity cover on home loans, a total that excludes job loss cover².

A right to leave does not change who sets the default. The default is still offered at the point of sale, by the party paid on that sale. Using the right later means gathering documents, comparing guarantees line by line and following a procedure. The incentive sits upstream. The work sits with the individual.

Better technology will not close this gap on its own. Every firm in the chain can adopt the same models. An insurer will deploy them to improve its risk selection and its margins. A distributor will deploy them to raise conversion. A comparison site will deploy them to grow its traffic. Each use is legitimate, and each inherits the objective of the business that runs it.

AI changes what is economically possible. It does not change whose interests a system is built to serve.

The common question is which product is best. It is hard to answer well, because “best” depends on the person, and any answer depends on who is doing the comparing. A more useful question comes first: for whom was this product optimized?

Seen this way, a commission rate, a claims ratio and a default choice are not technical details. They show what a product was built to serve.

Understanding one person’s situation, contracts and changes of circumstance, continuously, used to be too costly to do at scale. Products served as the scalable substitute. AI lowers that cost. An AI that works on the person’s side changes the function being optimized: the person’s interests become the objective, and products become the means. That is the premise behind Parley.

Parley is a broker too, paid by the insurer’s commission and by a fee. We publish both, so anyone can check whom Parley works for.