# Reading the connected vehicle: telemetry as a business asset
A modern connected car can generate several gigabytes of data per hour when you count camera, radar, and internal sensor streams. Even a conventional vehicle without advanced driver assistance pushes a steady flow of readings across its internal network: engine temperature, wheel speed, brake pressure, battery state, and hundreds of other signals, updated many times per second.
Most of that data never leaves the car. The strategic question for automotive companies is not "how much data exists?" It is "which signals are worth moving, storing, and acting on?"
This lesson traces where vehicle data comes from and how three business functions (warranty, insurance, and product) turn raw telemetry into decisions.
The CAN bus (Controller Area Network) is the internal messaging system that lets a car's electronic control units talk to each other. An ECU (Electronic Control Unit) is a small computer that runs one function: the engine, the transmission, the anti-lock brakes, and so on. A typical car has dozens of them.
CAN messages are compact. Each carries an ID and up to 8 bytes of payload. But they fire constantly, so the volume adds up. A raw CAN message might look like this:
Timestamp ID Data (hex)
1738000000.021 0x0C4 8A 01 00 00 00 00 00 00That 0x0C4 might mean "steering angle," and the bytes encode the value. Decoding requires a
Location data comes from the vehicle's GPS receiver, often fused with wheel speed and inertial sensors for accuracy in tunnels or cities. Position, speed, heading, and timestamp are small and cheap to transmit, which is why they are among the most widely collected signals.
Cameras, radar, ultrasonic sensors, and (on some vehicles) lidar produce the largest volumes by far. A single camera stream can dwarf all CAN and GPS data combined. This is why almost no company uploads raw sensor feeds continuously. Instead, they upload snippets: a few seconds around a hard braking event, or a frame flagged by an on-vehicle model.
A telematics control unit sits between the car and the cloud. It decides what to collect, buffers it, and transmits over cellular. Because bandwidth costs money and battery drains, this unit is where the "which signals matter" decision physically happens.
Warranty is often the first place data pays for itself.
When a component fails, the maker owes a repair. Multiply that across a fleet and small failure rates become large costs. Telemetry lets engineers see failures forming before claims spike.
Concrete example. Suppose battery temperature readings on a group of electric vehicles trend higher than the rest of the fleet under fast charging. Engineers can flag those vehicles, inspect the charging software, and issue an over the air update before failures cascade into claims. Catching a pattern in data is far cheaper than a recall.
Telemetry also fights warranty fraud and misdiagnosis. If a customer reports a transmission problem, logged data can show whether the fault actually occurred, how often, and under what conditions. That shortens diagnosis and reduces "no fault found" repairs, where a shop replaces a part that was never broken.
The key signals here are diagnostic trouble codes (standardized fault codes stored by ECUs, often called DTCs), component temperatures, duty cycles, and error counters.
Usage based insurance (UBI) prices coverage on how and how much someone actually drives, rather than on broad demographic proxies.
The relevant signals are behavioral: hard braking, rapid acceleration, cornering forces, speed relative to limits, time of day, and total mileage. Notably, most UBI programs do not need camera data. They run on cheap CAN and GPS signals, which is why UBI scaled faster than richer data products.
Concrete example. A driver who commutes 6 miles on quiet roads at midday is statistically lower risk than one who drives 40 highway miles at night with frequent hard braking. UBI lets an insurer see the difference and price it.
Two structural points matter for the business.
First, consent and privacy. Location and driving behavior are sensitive personal data. In the European Union, the GDPR (General Data Protection Regulation) requires a clear legal basis and transparency for processing it. In the United States, several state laws (for example the California Consumer Privacy Act) grant similar rights. In 2026, regulators and journalists continue to scrutinize how driving data flows from carmakers to insurers and data brokers. You can read the EU regulator guidance on connected vehicles from the European Data Protection Board.
Second, who owns the pipe. The carmaker controls the telematics unit, so it controls access. Insurers increasingly want that access, which creates both partnership opportunities and disputes over data rights.
Product teams use fleet telemetry to answer questions that lab testing cannot.
How do people actually use the car? Aggregated data reveals real world charging patterns, average trip lengths, how often a feature is engaged, and which driver assistance functions get switched off (a strong signal that something annoys users).
Where does the product fail in the field? Rare edge cases (a sensor confused by a specific road marking, a false alarm triggered by a certain sign) show up in fleet data long before they appear in structured testing.
Concrete example. If drivers repeatedly disable a lane keeping feature on a particular class of road, that is a product defect signal, even if nothing technically "broke." Engineers can locate the road types, reproduce the behavior, and tune the software.
This is the flywheel behind driver assistance development: fleet data surfaces edge cases, engineers improve the model, and updated software ships over the air. The vehicles that generate the most useful data are the ones already deployed.
Vérification des acquis
1. According to the lesson, what is the central strategic question automotive companies should ask about vehicle telemetry?
2. Why is raw CAN bus telemetry described as 'worthless' to a party that lacks the manufacturer's DBC file?
3. What best describes the role of an ECU in a vehicle's architecture?
4. Select ALL correct answers about the characteristics of CAN bus messages described in the lesson.
Sélectionnez toutes les réponses correctes.
5. Select ALL correct answers about how location data is handled in a connected vehicle.
Sélectionnez toutes les réponses correctes.
Data only becomes an asset when its value exceeds the cost of collecting, moving, storing, and governing it. Four principles separate the two.
Because bandwidth and storage cost money, the smartest programs process data on the vehicle and transmit summaries or triggered events. Sending "hard braking event at these coordinates, with 3 seconds of context" beats streaming everything.
Raw CAN data without the DBC mapping is noise. Whoever holds the signal definitions holds the value. This is a genuine competitive advantagecompetitive advantageA lasting edge over competitors: a resource, capability or position they cannot easily replicate, letting a firm earn above-average returns over time.Voir la définition complète → for vehicle makers and a friction point for third parties.
Consent, retention limits, and access controls are not just compliance overhead. A clean, well governed data supply chain is what lets a company sell insurance partnerships or share data with suppliers without legal exposure. Weak governance can shut a data productdata productA data asset managed like a product, with an owner, defined users, guaranteed quality, and measurable business value.Voir la définition complète → down overnight.
One car's data helps that one car. A million cars' data reveals failure patterns, usage trends, and rare edge cases that no single vehicle could show. The strategic asset is the fleet, not the vehicle.
For any signal, ask three questions:
1. What decision does it drive? (A warranty flag? A price? A software change?)
2. What is the cheapest form that supports that decision? (A summary usually beats a raw stream.)
3. What consent and governance does it require? (Behavioral and location data carry the heaviest obligations.)
If a signal drives no decision, it is a storage bill, not an asset.