How the Car Problem Lookup reports are made

In short

• The reports combine U.S. safety recalls and owner complaints from NHTSA, UK MOT test results, and UK and EU recall notices, for 19 brands, 302 models and model years 2012–2027.
• The database currently holds 7,650 U.S. recalls and 337,698 owner complaints across 2,188 model-year reports.
• No report text is written by an AI model or by hand for a specific car. Each report places official figures into the same layout, and quotes owners in their own words with the public complaint number, so every claim can be checked at the source.
• Complaints are not verified and do not rate reliability. They show what owners reported, which is most useful for spotting patterns and comparing model years.

Where does the data come from?

Each report shows the date its data was retrieved. Before any refresh is published, an automatic check compares the new totals with the previous ones; if recalls, complaints, models or pages drop by more than 2%, which would point to an outage at the source rather than real news, the refresh is not published.

Which cars get a report?

NHTSA lists vehicles under the exact names manufacturers file, such as "SILVERADO 1500" or "328I XDRIVE". These are grouped into the model families people search for (Silverado, 3 Series, GLC-Class), and owner complaints and recalls are de-duplicated within each family and model year. Commercial chassis, motorcycles and heavy trucks are left out.

• Published: a model year gets a report when NHTSA has at least one recall or at least three owner complaints for it.
• Shown in search engines: a report is offered to search engines when it has at least 10 owner complaints or UK MOT data for that year. Reports with fewer reports stay available to readers on CarHooq but ask search engines not to list them, because a handful of complaints is too little to be useful on its own. They are listed automatically once enough reports arrive.

How are the symptoms and owner questions found?

The text of every owner complaint is checked against a list of about 50 well-defined symptoms, such as stalling or loss of power, a check engine light, harsh transmission shifting, unexpected automatic braking, oil consumption or a shattering sunroof. Each symptom is described by precise word patterns, which were checked against a sample of more than 190,000 complaints to confirm they match what owners actually describe.

• A symptom appears on a report only when at least three owners and at least 3% of that model year's complaints describe it.
• Each answer contains only: how many owners described it, their share of the complaints, a related recall if one matches the same part, and a short quote from one owner with its NHTSA complaint number.
• Questions are phrased the way owners search ("Why does my 2019 Volkswagen Tiguan stall, lose power or shut off while driving?"), but the answer never adds causes, diagnoses or repair advice that the data does not contain.

How are the owner quotes chosen?

Each report shows up to 12 complaints: the most recent ones for the components owners mention most, so the quotes reflect the main problem areas rather than the most dramatic stories. The text is exactly as filed with NHTSA, which removes names and VINs; long reports are trimmed with an ellipsis. CarHooq does not edit, verify or select complaints for tone.

How are the UK MOT failure rates calculated?

• Only class 4 cars and normal MOT tests are counted. A test counts as a fail if the car failed, including cars repaired and passed within an hour at the test station (PRS).
• MOT data is grouped by the year a car was first registered, which is usually, but not always, its model year.
• Model names in MOT data are typed by testers, so they are mapped to the same model families as the U.S. data. Cars sold under different names in the UK (Ford Kuga and Escape, for example) are matched; different cars that share a name are not.
• Any group with fewer than 100 tests is not shown. "Points better or worse than average" compares a model with all cars of the same age tested that year.
• Failure areas show the share of failed tests with at least one fault in that area; a test can fail in several areas.

How are UK and EU recalls matched to a car?

UK and EU recall notices name models and give build or production dates. These are matched to the model families above, and the build dates are turned into the model years they cover. When a notice gives no dates, the affected years are estimated from the recall date and the recall is labelled "Years estimated". European models can differ from U.S. models with the same name, so readers should always confirm with their registration number or VIN.

What can the numbers not tell you?

• Complaints are not verified. NHTSA publishes them as submitted; a single complaint proves little.
• Counts follow sales. Popular cars collect more complaints because more of them are on the road. Compare a car with its rivals and with its own other model years.
• Newer model years look better than they are. They have had less time to collect complaints, so CarHooq leaves the newest years out when it names the years with the fewest complaints.
• Model-year recalls are not your car's status. A report lists recalls for the model year, not whether a particular car has been repaired. Check a VIN at nhtsa.gov/recalls or a registration at check-vehicle-recalls.service.gov.uk.
• Not mechanical advice. Reports describe what owners and regulators recorded. For a diagnosis, see a qualified technician.

Who runs CarHooq?

CarHooq is run by Rahul Wagh, a software engineer and Mercedes-Benz owner (a 2021 GLC 300 plug-in hybrid), not a mechanic. The Car Problem Lookup was built so owners and used-car buyers can read official recall and complaint data without digging through government databases. CarHooq is independent: it is not affiliated with any carmaker, dealer or government agency, and it is supported by advertising, which never affects what a report says.

Found an error, such as a recall matched to the wrong model? Please tell us and it will be corrected in the next refresh.