Every motor premium is a bet on two numbers: how often cars like yours crash, and how much each crash costs. Understanding the split explains most of what looks mysterious about pricing.
Strip away the rating factors, the discounts and the actuarial vocabulary, and a motor premium is built from two numbers multiplied together: the probability that a policy produces a claim in a year, and the average cost of a claim when it happens. Frequency times severity gives the expected claims cost per policy — the burn cost — and everything else in the premium is loading on top of that for expenses, reinsurance, capital and margin.
This piece is a plain-language tour of those two curves: what moves each one, why they often move in opposite directions, and why anyone distributing motor insurance — a marketplace, a fleet platform, a dealer — should care about the split and not just the headline price.
Frequency: how often the phone rings
Claims frequency is usually expressed as claims per policy per year. It is a behaviour curve more than a vehicle curve — it responds to how much driving happens and in what conditions.
- Exposure. Kilometres driven is the master variable. A car parked most of the week generates a fraction of the claims a daily commuter does, which is the entire logic behind usage-based products.
- Environment. Dense urban traffic produces frequent low-speed contact; highway driving produces rarer, harder impacts. City frequency is high; city severity, per claim, is often modest.
- Driver mix. Age, experience and claims history are frequency predictors everywhere motor insurance is priced. This is why a change in who buys through a channel changes that channel's loss experience even when the cars are identical.
- Systemic shifts. Anything that changes driving across a whole market moves the curve for everyone at once. Fuel prices, remote work, road enforcement, even weather patterns show up in market-wide frequency before any insurer's individual data explains why.
Severity: how big the bill is
Severity is the average cost per claim, and it is largely an economy curve. The car crashes in traffic; the claim is settled in the repair market.
- Parts and labour. Repair cost inflation flows straight into severity. A market importing most spare parts inherits exchange rates and shipping costs inside its claims.
- Vehicle technology. A bumper with sensors and cameras in it costs multiples of a plain one. Safety technology can cut frequency and raise severity simultaneously — fewer crashes, dearer ones — which is the single best illustration of why the curves must be watched separately.
- Injury and liability costs. Bodily injury claims sit at the far end of the severity distribution, and legal or compensation frameworks set their scale. A small number of large injury claims can matter more to the total than thousands of fender repairs.
- Total-loss thresholds. When repair costs rise faster than used-car values, more damaged cars tip over the write-off line. That reshapes the severity distribution rather than just shifting its average.
Why the split matters more than the total
Two portfolios can have identical burn costs and completely different risks. A high-frequency, low-severity book — urban commuters, minor collisions — is statistically stable: many small claims average out, and pricing can be tuned with confidence. A low-frequency, high-severity book is jumpier: results swing on a handful of large losses, and a good year proves less than it seems to.
A premium is not a price for a product. It is a forecast of two curves, with a margin for being wrong.
For a platform distributing motor cover, three practical consequences follow.
- Repricing has a direction you can anticipate. If your market's claims inflation is running hot, expect severity-driven premium increases at panel renewals regardless of how well your particular customers drive. Build forecasts on attach rates and volume, not on premiums holding still.
- Your channel changes the curves. Embedded distribution reaches customers at a purchase moment, which selects differently than aggregator shopping does. Insurers will eventually price your channel on its own experience; a platform that passes clean vehicle and driver data speeds that up in its own favour when the experience is good.
- Product design is curve design. Deductibles trim frequency by absorbing the smallest claims. Agency-repair options raise severity by fixing where the car gets repaired. Every product lever you offer in a checkout is quietly a bet on one curve or the other.
The honest caveats
This explainer simplifies on purpose. Real pricing works on distributions rather than averages, severity's long tail resists averaging, and the two curves are not independent — the same congestion that raises frequency suppresses speed and severity. Nothing here uses market-specific figures, because frequency and severity levels vary sharply by country, line and year; for the Saudi numbers, the Insurance Authority's annual market disclosures are the place to look, and our market-data pieces on this blog track them. The framework, though, travels everywhere motor insurance is sold: when your premium moves, the first question worth asking is always which curve moved it.