THE CHALLENGEPricing on How People Drive, Beyond Who They Are
Traditional motor pricing leans on proxies like age and postcode that say little about how a given person actually drives. Insurers want to price on real behaviour, reward safe drivers, and detect crashes the moment they happen, but the telematics under that promise is hard. It has to onboard customers with almost no friction, score driving in a way the actuaries trust, catch a real crash without crying wolf on every pothole, and feed clean data into policy and claims systems. The full stack covers all of that so the insurance team can focus on the product, not the plumbing.
Part of the Telematics and GPS Tracking stack, and commonly built alongside Connected Vehicle and OEM Telematics.
WHAT'S INCLUDED
Hardware, App, and Backend for UBI
OBD and App Data Capture
OBD-II dongles capture rich vehicle data, and an SDK and app cover phone-only programs. The app captures trips from GPS and the phone IMU, the dongle adds true odometer, engine state, and DTCs, and many programs run a hybrid of both.
Driver Scoring Model
Each trip is scored on harsh braking, acceleration, cornering, speeding against posted limits, distraction, and mileage and time-of-day exposure, then rolled into a driver score. The weights are tuned with your actuarial team so the score maps to your pricing.
Harsh-Event Detection
Harsh braking, acceleration, and cornering are detected on-device from IMU thresholds calibrated per vehicle class, so events are consistent across phones and dongles and not skewed by where the phone sits in the car.
Crash Detection
A high-rate IMU samples fast enough to catch an impact signature, combined with speed change and the vehicle bus on a dongle to suppress false positives. A confirmed crash triggers first-notice-of-loss and an optional policyholder check-in.
Insurer API Integration
Scores, trip summaries, mileage, and crash events are exposed over REST and webhooks and mapped to your policy administration and claims platforms, feeding pricing at renewal and opening claims automatically on a confirmed crash.
Policyholder App
The customer-facing app carries trip history and maps, a driver score with feedback, earned discounts and rewards tied to your program, and a clear crash and assistance flow that keeps customers engaged and enrolled.
CRASH DETECTION
Catching the Impact Without Crying Wolf
Crash detection is the feature most likely to fail in the field, either missing real impacts or firing on every dropped phone. It is treated as a signal-processing problem, not a single threshold.
High-Rate Sampling
A normal trip logger samples too slowly to see a crash. The IMU runs at a high rate during motion so the short, sharp acceleration spike of an impact is captured with the shape needed to tell it apart from ordinary driving.
Multi-Signal Confirmation
The acceleration profile is combined with the speed change before and after and, on a dongle, the vehicle bus. A real crash shows all three together, which suppresses false positives from a dropped phone, a pothole, or a kerb strike.
Response Workflow
A confirmed event opens a first-notice-of-loss record, pushes a check-in to the policyholder app, and can escalate to assistance. Severity is graded so a minor bump and a major collision route to different workflows.
ARCHITECTURE
From Trip to Pricing Engine
Edge and App
The OBD dongle runs firmware on an STM32 with a GNSS receiver, a high-rate IMU, and a cellular or BLE link to the phone. The mobile SDK handles trip detection, scoring features, and crash signals on Android and iOS with low battery impact.
Ingestion and Scoring
Trips publish over MQTT and HTTPS into an ingestion pipeline that validates, deduplicates, and scores them. A Kafka stream feeds the scoring service and writes trips and events into a time-series store for analytics and model tuning.
Insurer Integration Layer
A documented REST API and webhooks expose scores, mileage, and crash events to your policy, pricing, and claims systems, with role-based access so what the insurer sees is separated from raw policyholder location data.
PRIVACY AND COMPLIANCE
Built for Trust and Regulation
Transparent Consent
The policyholder app shows drivers their own trips and score and states what is collected and why. Clear consent and visible value raise enrolment quality and keep customers in the program rather than opting out at renewal.
Data Protection
Location and trip data are encrypted in transit and at rest, retention windows are configurable to your regulatory regime, and access is segmented so the insurer side sees scores and summaries rather than raw position trails by default.
Actuarial Alignment
Score weights are tuned alongside your actuarial and pricing teams and validated against your claims data, so the model is defensible to regulators and maps to real loss experience rather than a generic index.
FAQ
Common Questions
OBD dongle or smartphone app: which data source fits?
Both work and both are supported. A smartphone app gives the lowest-friction onboarding with no hardware to ship, capturing trips from the phone GPS and IMU. An OBD dongle gives richer vehicle data such as actual odometer, engine state, and DTCs, and removes the dependence on the policyholder carrying their phone. Many programs run a hybrid, app-first with an optional dongle for higher-value policies.
How is the driver score calculated?
The scoring model is built around the behaviours that correlate with claims: harsh braking and acceleration, cornering, speeding against posted limits, phone distraction where the app can detect it, and time-of-day and mileage exposure. Each trip is scored and weighted into a rolling driver score. The weights are tuned with your actuarial team so the score maps to your pricing model rather than a generic index.
How reliable is the crash detection?
Crash detection runs off a high-rate IMU, sampling acceleration fast enough to catch the impact signature a normal trip logger would miss. The acceleration profile is combined with speed change and, on a dongle, the vehicle bus to cut false positives from drops and potholes. A confirmed event triggers a first-notice-of-loss workflow and an optional check-in with the policyholder.
Can it integrate with your policy and claims systems?
Yes. Driver scores, trip summaries, mileage, and crash events are exposed over a documented REST API and webhooks, and mapped to your policy administration and claims platforms. Designed to feed pricing engines at renewal and open first-notice-of-loss records automatically on a confirmed crash.
How do you handle data privacy and policyholder trust?
Transparency is built in. The policyholder app shows drivers their own trips and score and explains what is collected and why, which raises engagement and consent quality. Location and trip data are encrypted in transit and at rest, retention is configurable to your regulatory regime, and you control what the insurer side sees versus what stays on the device.
What does the policyholder app give the end customer?
A trip history with maps, a driver score with feedback on what to improve, any earned discount or reward tied to your program, and a clear crash and assistance flow. Giving drivers visibility of their own behaviour is what turns a tracking program into one customers stay enrolled in.
Ready to Launch Your UBI Program?
Share your product, your data sources, and your policy and claims stack to get a tailored approach across onboarding, the scoring model, crash detection, and the insurer integration.
Schedule a Free Consultation