The risk is real. It is driving. You just cannot see it yet.
Across Central and Eastern Europe, motor insurers are watching the same thing happen at the same time. Claims costs are climbing while premiums lag. Combined ratios are drifting in the wrong direction, and on MTPL the drift is now structural in several markets.
Nothing dramatic has changed in underlying accident risk. What has changed is the gap between what the pricing model can see and what is actually happening on the road.
That gap is where the margin is going.
What the numbers say
In Poland, motor insurance payouts hit a record €4.4 billion (PLN 19.24 bn) in 2024 — up roughly PLN 3 bn year on year (XPRIMM). MTPL claims grew 12.6% and motor own-damage claims 18.2% in 2024 (Polish Chamber of Insurance, summarised by Czublun), while average MTPL premiums rose just 7.0% through 9M 2024 (VIG 2024 Annual Report — Poland segment). The structural gap between claims inflation and premium catch-up keeps widening.
In Slovakia, MTPL is running at a 117% combined ratio as of 30 June 2024 — figure attributed to the National Bank of Slovakia and cited by ČSOB Poisťovňa in its Annual Report 2024. Every euro of premium pays out €1.17 in claims and costs. The line is structurally unprofitable, and price competition keeps pushing premiums down.
In Romania, the government extended the MTPL premium cap through March 2025, freezing pricing flexibility while costs continue to climb (CMS Law-Now, December 2024). MTPL claims paid rose 36% in 9M 2024 (XPRIMM). The cap stops insurers raising rates; it does not stop bodily-injury inflation.
In the Czech Republic, VIG’s segment combined ratio deteriorated to 94.8% in 2024 from 91.3% the year before — a usable proxy for market direction (VIG 2024 Annual Report — Czech segment). In Slovenia, the industry net combined ratio sat at 98% in 2024, improved from 102% in 2023, close enough to breakeven that pricing discipline matters (Triglav Re Annual Report 2024). In Croatia, motor lines accounted for 61% of all non-life claims paid in 2024 (XPRIMM), making motor the central pressure line in the market. Hungary and Bulgaria show the same direction with less public granularity at the line-by-line level.
Different markets, same story. The cost of risk is rising faster than the price of risk.
Why static pricing cannot keep up
Traditional motor pricing is built on proxies. Age, vehicle, postcode, claims history, sometimes mileage. These tell you who someone is. They tell you almost nothing about what they do once they are behind the wheel.
A careful, low-mileage 28-year-old who drives mostly on motorways looks identical, on paper, to a reckless 28-year-old who drives heavily at night on rural roads. They get priced the same. One subsidises the other until the claim arrives, and by the time the claim arrives the pricing decision is months or years old.
The pricing model is doing its job. The data feeding it just isn’t enough.
The adverse-selection trap
Price competition in CEE is increasingly run through comparison portals and multi-agent platforms. That changes the dynamics in a way many actuaries have stopped pretending is benign.
Insurers competing on price attract a disproportionate share of high-risk policyholders. Good drivers, the predictable and profitable ones, shop around and leave for better-priced offers. Bad drivers, who know they are getting a favourable deal relative to their actual risk, stay put.
The portfolio quietly deteriorates. No reinsurer alert, no regulator letter, no underwriter waving a flag. Just a loss ratio that drifts the wrong way, quarter after quarter.
Aggressive static-data pricing wins the wrong customers. The worst risk lands at the cheapest price, and the good drivers walk anyway.
What an answer would need to look like
The obvious counter to invisible risk is to find a way to see it. The piece of information static pricing is missing is the one that actually predicts claims: how the policyholder drives. Behavioural data is the missing input — how often, when, on what kind of road, with how much smoothness or attention.
Collecting that data at portfolio scale means putting a telematics product in front of the driver. For years that meant a hardware box wired into the vehicle; today it almost always means an app on the driver’s phone, either standalone or as scoring code embedded in the insurer’s own app via an SDK.
But before any product like that gets serious traction in CEE, it has to clear four bars. The same four come up in conversations with CUOs, Heads of Motor and pricing leads across the region:
- Loss-ratio performance is the metric that matters. Features are noise.
- ROI has to be quantifiable before the pilot, not after the rollout.
- Implementation complexity is the silent killer of good ideas. Anything that needs device shipment, vehicle installation or a long IT roadmap gets deprioritised.
- Regulatory and data risk are non-negotiable. ISO 27001 and GDPR-by-design are entry conditions, not differentiators.
Every one of those is a business-confidence question. Can we measure the impact, can we afford the implementation, and can we live with the data risk? Technology comes after the answers to those.
Behavioural data, in practice
In practice, behavioural data is a handful of signals validated against real claims data: frequency, timing, road type, smoothness of acceleration and braking, attention. Insurers call this exposure to risk — measured continuously, not estimated up front.
Behaviour-based pricing has been live in mature Western European markets for years. What has changed is the cost of access.
App-based telematics has removed the hardware barrier. No OBD dongles, no installation logistics, no warehouse full of devices waiting to be shipped. The smartphone in the policyholder’s pocket already collects most of what matters; the rest is signal processing.
Berg Insight tracks 13.8 million active insurance telematics policies in Europe at end-2024, growing at a 7.8% CAGR toward 20.1 million by 2029 (Berg Insight, Insurance Telematics in Europe and North America, 9th edition, December 2025). Poland is the CEE leader, with over 75% of insurers having tested or implemented telematics in some form (Wybierz-ubezpieczenie, June 2025). The rest of the region is several years behind, which is precisely the opportunity.
The MOVE Score: behavioural risk, delivered as data
Dolphin’s MOVE Score is a behavioural risk score validated on more than 62,000 drivers across roughly 310,000 driver-years of observation against actual MTPL and CASCO claims (MOVE Score whitepaper, 2025). It is delivered via API. The insurer queries the score; the driver uses the MOVE Score app.
The validation surfaces specific, priceable patterns. 25% of all accidents happen in the first three minutes of a trip. Trips over 40 minutes carry 2.5× the risk of shorter ones. Rural roads run at 2.14× highway risk; urban roads at 1.92×. Higher MOVE Scores correlate with measurably lower claim frequencies on both MTPL and CASCO.
For an insurer who wants to test the impact of behavioural data on pricing and underwriting, this is a low-commitment way in. No app build. No IT-roadmap rewrite. The score plugs into the existing underwriting model the same way any external data feed does, so you can validate the business case before committing to anything bigger.
For insurers ready to go further, the same data foundation underpins two other entry points:
- MOVE SDK — for insurers who already have or plan to build a customer-facing app and want to embed scoring, trip data and engagement features natively.
- White Label App — Dolphin builds the full branded product: driving scores, rewards, challenges, push notifications, customer engagement. The insurer owns it. Live examples: Generali Mobility App, Porsche Smart Driver App, UNION Drivello App.
Start with a score on an API, prove the case, and step up only when the numbers justify it.
What changes when behaviour is in the model
Three things move when a behavioural layer is added to motor pricing.
Pricing accuracy improves, because the insurer can finally distinguish between policyholders who look identical on paper but behave very differently on the road. Risk gets priced where it actually lives.
Adverse selection reverses. Good drivers, the ones the portfolio needs, see a price that reflects how they actually drive. They have a reason to stay. High-risk drivers are priced correctly. The portfolio rebalances without changing the average market premium.
Loss ratios respond. Better pricing and better selection compound across the renewal cycle. The point: prices land where the risk actually sits, instead of being averaged across people who don’t share it. Average premium doesn’t have to move.
Privacy as a precondition, not a feature
One reason CEE adoption has been cautious is that privacy and regulatory exposure feel disproportionate to the upside. That is fair, and it should be the default position.
The MOVE Score is delivered as a score. Only the score crosses to the insurer. Location traces and raw behavioural signals stay encrypted and stay with the driver. ISO 27001 certified. GDPR-compliant by design.
That’s an architectural choice, baked into how the product is built. It’s also why Generali, Vienna Insurance Group, Kooperativa, Porsche Insurance and UNION Biztosító run their telematics programmes on Dolphin’s stack.
Where to start
The honest answer to “what do we do about loss-ratio pressure”:
- Quantify the cost of inaction in your own portfolio. How much of the 2024 loss-ratio drift is structural pricing, versus claims inflation you cannot price around?
- Run a small pilot with behavioural data alongside the existing model. Compare the score’s predictive power against your current rating factors on real claims data. Dolphin’s MOVE Pilot programme is built for exactly this.
- Decide based on the result, not the deck.
Motor insurance risk has always been about driving. What has finally changed is that we can see it.
If your loss ratio looks worse than it should this year, the issue is probably visibility, not risk. The portfolio is the same shape it was last year. Your model just sees less of it.
