Usage-Based Car Insurance: How Telematics is Rewriting the Underwriting Rulebook

What Telematics Actually Measures

The data inputs that feed a usage-based insurance programme fall into several categories, and understanding them is the prerequisite for understanding what the resulting premium actually reflects.

The simplest category is mileage. Pay-per-mile insurance — sometimes called Pay-As-You-Drive — charges a base rate plus a variable rate per mile driven, measured directly through a tracking device. If you drive 4,000 miles a year, you pay considerably less than someone driving 16,000 miles in the same vehicle, because miles driven is the single variable most directly correlated with accident exposure. This model is straightforward, relatively transparent, and the privacy implication is modest: the insurer learns how much you drive and roughly where you go.

Pay-How-You-Drive goes further, measuring not just distance but behaviour. Usage-based auto insurance telematics collects data on mileage, speed, braking, acceleration, and time of day via OBD-II devices, smartphone apps, or embedded systems — with primary types encompassing pay-as-you-drive, pay-how-you-drive, pay-per-mile, and hybrid approaches. The behavioural variables have a strong actuarial basis: hard braking events correlate with following too closely or inattentive driving; rapid acceleration correlates with aggressive driving styles; late-night driving correlates with fatigue and elevated accident risk in population data. These correlations are real and statistically meaningful, which is why they produce more accurate risk pricing than demographic proxies.

The three primary hardware delivery systems are the OBD-II plug-in device, the smartphone app, and embedded factory telematics. OBD-II devices led the market with 34.69% revenue share in 2025, while smartphone-based systems are growing fastest at a 14.05% compound annual growth rate through 2031, because smartphone telematics eliminates installation costs and uses native gyroscopic and GPS sensors, making it accessible without any physical device. Embedded telematics — factory-installed systems that the automaker controls — represents the fastest-growing category because they remove the consumer opt-in decision almost entirely; the capability is built into the vehicle at manufacture.

The National Highway Traffic Safety Administration has published Cybersecurity Best Practices for the Safety of Modern Vehicles, addressing the security obligations of connected vehicle systems — including telematics infrastructure — and the standards that manufacturers and their data partners are expected to meet for protecting data transmitted from vehicles. The NHTSA guidance reflects federal recognition that vehicle data is not merely commercial but safety-critical, and that the infrastructure transmitting it requires security standards comparable to other critical infrastructure.

The Actuarial Logic Behind Behaviour-Based Pricing

To understand why usage-based insurance represents a genuine shift in underwriting philosophy rather than simply a marketing differentiator, it helps to understand what traditional actuarial models are doing and where they fall short.

Traditional auto insurance pricing assigns you to a risk segment defined by observable characteristics that correlate with claims frequency in historical data. Your age group, your zip code, your vehicle model, your claims history, and your credit score are all variables that predict, at a population level, how likely drivers like you are to file claims. The model works in aggregate — the correlation between these variables and claims frequency is real — but it applies group averages to individuals who may differ significantly from their statistical cohort. A 22-year-old who drives cautiously, rarely, and exclusively in daylight pays the same rate structure as a 22-year-old who drives aggressively, frequently, and after midnight, because the traditional model cannot distinguish them.

The National Association of Insurance Commissioners — NAIC — has described usage-based auto insurance as tracking driving behaviors via telematics to set premiums, with the benefit of incentivising safer driving while raising privacy concerns, and the acknowledged limitation that the NAIC notes “not everyone is a better than average driver” — some participants end up paying more than they would have under a traditional policy. That last observation from the NAIC is the honest counterweight to the “fairness” pitch that UBI marketing tends to lead with. The individualisation that UBI enables cuts in both directions. For drivers who are genuinely safer than their demographic average, UBI produces lower premiums and is experienced as fair. For drivers who are worse than their demographic average, UBI produces higher premiums and is experienced as punitive — even though it is, by the actuary’s standard, more accurate.

The scale of adoption indicates that the market believes the efficiency gains are worth the complexity. As of 2024–2026, approximately 20% of US auto insurance policies incorporate some form of telematics, up from roughly 15% in 2022 and less than 10% in 2020, representing 45–50 million active telematics policies in the United States. The growth trajectory indicates that insurers are finding the model commercially useful — lower-risk drivers are self-selecting into UBI programmes, which improves the portfolio quality of the carriers offering them.

The AI layer that has been added to telematics processing in 2026 extends the model significantly. New tools such as advanced driver monitoring, contextual risk scoring, and incident detection are now embedded in hybrid telematics systems that are improving underwriting accuracy further, with AI-powered risk engines and behavioural trend analysis combining to deliver a risk assessment capability that static historical models cannot approach. The practical consequence for policyholders is that the model assessing your risk is continuously learning — not just averaging your behaviour over a programme period, but detecting patterns and anomalies that a static scoring approach would miss. How those inferences translate into pricing decisions is not fully transparent to policyholders, which is where the regulatory conversation is increasingly focused.

Your Data, Their Decisions: The Rights You Should Know

The data governance dimension of telematics insurance is where consumer rights are most consequential and most underexplained. When you enrol in a UBI programme — whether through an OBD-II device, a smartphone app, or your connected vehicle’s native telematics — you are consenting to data collection under terms that deserve careful reading before you sign.

The Fair Credit Reporting Act governs the use of telematics data in insurance rating and policy management — sharing of driving data is subject to FCRA oversight, and insurers are required to disclose how they use driving information and how it affects your premium. Insurers need clear, informed consumer consent before collecting or using telematics data, and data should only be used for insurance rating and policy management. Those are the requirements as they exist on paper. The consent you provide through a programme enrolment agreement can, however, be considerably broader than those baseline requirements suggest.

The FTC has finalised an order settling allegations that GM and OnStar collected and sold geolocation data without consumers’ meaningful consent — a case that established that vehicle telematics data shared with third-party data brokers without adequate disclosure can constitute an unfair or deceptive practice under Section 5 of the FTC Act. The GM case is the most significant federal enforcement action on vehicle data privacy to date, and it sends a signal to insurers and automakers that telematics data sharing practices are under active FTC scrutiny. It does not mean that all telematics data sharing is prohibited — it means that sharing must be meaningfully disclosed to and consented to by the driver.

The Supreme Court’s 2018 decision in Carpenter v. United States — in which the Court held that the government requires a warrant to access historical cellphone location data because it constitutes a detailed record of physical movements — established a constitutional principle with implications for how granular telematics data is treated in legal contexts. While Carpenter specifically addressed government access rather than commercial data sharing, it reflects a broader judicial recognition that persistent location tracking produces a uniquely intimate portrait of a person’s life that deserves heightened protection.

As of 2025, roughly twenty states have enacted comprehensive consumer data privacy laws, many of which classify precise geolocation data as sensitive personal information with additional protection requirements — and some participation agreements allow the insurer to share driving data with affiliated companies or data analytics partners, with previously collected data not automatically deleted if a consumer leaves the programme. That last point is the one most people don’t think to check before enrolling: leaving a telematics programme does not necessarily erase the data already collected. Understanding the data retention terms before you enrol is not paranoia. It is basic consumer due diligence.

The deeper context of how telematics data flows through the broader insurance data ecosystem — and what rights consumers have to inspect and challenge the consumer reports built from it — is examined in our piece on the data rights economy and who controls the information that determines your insurance premium. The LexisNexis consumer disclosure report that contains your driving history data is an FCRA-protected document you can request annually and dispute for errors — a right that most UBI participants don’t know they have.

The Fairness Question the Industry Is Still Navigating

The NAIC began formally pressing for accountability on telematics data in 2025. In January 2025, consumer advocates asked the NAIC Property and Casualty Committee to provide guidance, gather telematics data, and publish a detailed report on usage-based insurance — reflecting growing regulatory concern that the expanding role of telematics in insurance pricing is outpacing the oversight frameworks designed to ensure those prices are fair and non-discriminatory. The NAIC’s comprehensive study on UBI — published as part of its insurance market regulatory series — acknowledged the efficiency benefits of telematics while raising the question that sits at the heart of the fairness debate: whether behaviour-based pricing can produce discriminatory outcomes through variables that correlate with protected characteristics, even when the variables themselves are facially neutral.

The concern is specific and documented. Night driving frequency, which correlates with accident risk, also correlates with shift work employment patterns that are unevenly distributed across racial and income lines. Sudden braking frequency, which correlates with aggressive driving, can also reflect driving in dense urban traffic conditions where hard braking is sometimes unavoidable rather than indicative of driver behaviour. The model cannot always distinguish between a driver who brakes aggressively because they drive aggressively and a driver who brakes frequently because they commute through downtown traffic. Our analysis of AI bias in insurance and whether algorithmic pricing can discriminate against consumers maps the proxy discrimination mechanism that makes this concern genuine and regulatory, not just theoretical.

The NAIC’s comprehensive study on Usage-Based Insurance and Vehicle Telematics: Insurance Market and Regulatory Implications examines how telematics changes the underwriting model, the data challenges involved, and the implications for insurers, consumers, and state regulators — establishing the foundational regulatory framework within which UBI programmes operate. For regulators and consumers alike, that study remains the primary reference document for understanding what UBI programmes are required to disclose, how state commissioners are approaching them, and what redress mechanisms exist when pricing produces outcomes that feel unfair.

The mechanism connecting telematics data to insurance pricing decisions — and what AI underwriting models do with the raw data before it becomes a premium — is explored in detail in our piece on how AI underwriting algorithms are setting insurance premiums in 2026. The short version is that the raw telematics data doesn’t directly become your premium; it passes through a scoring model that weights different behavioural signals according to their actuarial relationship with claims frequency, and those weights are proprietary. You receive a score. The scoring methodology stays with the insurer.

Frequently Asked Questions