Methodology & Data Provenance

Every figure in this report is tagged by what kind of claim it is. Where a variable is missing, it is named as missing rather than estimated or inferred.

Primary sources and why there are two of them

A survey of cars under five years old largely measures how a manufacturer behaves while it is still paying the bills. Warranty-claim data measures what happens once it stops. Using only one of these produces a materially different answer.

The Finding: The Velar Has Two Reliability Records

For four of these five models, the two datasets broadly agree. For the Range Rover Velar they contradict each other, and the contradiction is the single most useful thing in this report for anyone buying used.

Land Rover holds four of the ten lowest places in that index — Discovery (17.2), Velar (19.7), Range Rover Sport (21.1) and Discovery Sport (24.1), with Defender 110 at 28.0. Across that cohort, repair requests cluster at six to seven years old and just under 70,000 miles, emissions-related faults are the most common repair category affecting four of the ten models, and the group average repair cost is £2,051.63. [M]

The Audi Q5, BMW X3, Mercedes-Benz GLC and Volvo XC60 do not appear anywhere in that bottom ten. That absence is a finding in itself. [M]

The Index: Composite Scores & Valuation Modifiers

Where What Car? splits a model by fuel type, the petrol/PHEV derivative is used as the London-representative figure, on the basis that London’s used demand is skewing away from diesel. Both derivatives appear in section 04.

Composite and SAV modifier are London Auto Index calculations, reproducible from section 06. For context, What Car? gives the premium SUV class an average of 89.3%, the lowest of any class it rates — a separate figure on a different scale, and not an input here.[3]

The Fuel Split Changes the Answer

Treating each of these as a single entity hides the most actionable finding in the dataset. All figures below from the What Car? premium SUV results. [M]

Across the whole survey, What Car? found petrol cars had the lowest average fault rate at 19%, hybrids 21%, electric vehicles 27% and diesels 29%, with diesels also least likely to be repaired free of charge — 87%, against 92% petrol, 96% hybrid and 97% electric. [M]

Vehicle Intelligence: Model by Model

Mercedes-Benz GLC — composite 85.4

Faults. 21% of GLCs reported on had a problem — 12-volt battery 5%, non-engine electrics 4%, and air-con, hybrid electric motor, engine electrics, interior trim and infotainment at 2% each. Almost no powertrain content; the mildest mechanical profile of the five. [M]

Warranty conduct. Mercedes covered the cost of all repairs, but 44% of faulty cars sat in the workshop for more than a week. Strong on cost, weak on downtime. [M]

The caveat. The model performs well; the brand does not. Honest John readers placed Mercedes fifth-worst for reliability on 86.03%, with the C-Class — the GLC’s saloon sibling — fourth-worst among individual models on 83.66%. What Car? separately found 43% of post-2021 C-Class cars had a fault, against 23% for the previous generation. Publishing the model win without the brand context would be selective quotation. [M]

Volvo XC60 PHEV — composite 81.5

Faults. 23% fault rate for PHEVs against 28% for petrol and diesel. Infotainment dominates at 13%, then non-engine electrics 7%, air-con 3%, bodywork 3%, interior trim 3%, and electric motor, motor electrics, exhaust and suspension at 2% each. The most infotainment-concentrated profile here; mechanically close to clean. [M]

Warranty conduct — best in group. Volvo paid for all repairs, and PHEVs returned to the road faster than petrol and diesel cars — 36% off road for more than a week against 50%. [M]

The inversion. The plug-in hybrid is the more reliable XC60, running counter to the wider pattern — Consumer Reports finds PHEV variants generally score worse than their petrol-only equivalents, listing the XC60 PHEV among them in its US data. UK owner data points the other way. The two markets disagree, and that is worth knowing rather than resolving. [M]

BMW X3 petrol/PHEV — composite 78.8

Faults. 31% of petrol and PHEV X3s went wrong — bodywork 12%, infotainment 8%, and engine, exhaust, interior trim, non-engine electrics and suspension at 4% each. Bodywork dominance is unusual, and matters for valuation: a panel or paint fault is a very different asset risk from a gearbox fault. [M]

Warranty conduct. BMW covered all costs on petrol and PHEV models, but two-thirds of cars took more than a week to put right — the worst downtime in this group. On diesels, a small proportion of owners funded bills exceeding £1,500 per fault, though repairs were quicker. [M]

Context. The X3 is absent from the Warrantywise bottom ten, but the BMW X5 (23.6) and X7 (26.5) both appear in it. BMW’s larger SUVs age worse than the X3 does. [M]

Audi Q5 petrol/PHEV — composite 73.1

Faults. 39% of Q5 petrol and PHEV cars had an issue — the highest fault rate of the five: non-engine electrics 13%, engine/motor electrics 10%, air-con 7%, infotainment 7%, and battery pack, bodywork, electric motor and steering at 3% each. [M]

Warranty conduct — weakest here. The only model in this group where owners paid. Audi did not cover all remedial work, leaving some owners with bills of up to £750, and more than half of faulty cars were off the road for more than a week. 80% of repairs cost nothing; 20% cost £201–£750. [M]

The brand pattern. Audi’s cost-coverage behaviour, rather than its fault rate, is the variable to watch. The Q7 is the lowest-scoring premium SUV in the survey at 78.6%: 47% of cars went wrong, only 15% were fixed free, and 29% of owners received bills over £1,500 — against all faulty Mercedes GLBs and 90% of Volvo XC90s repaired at no cost. [M]

Range Rover Velar — composite 67.2

In warranty. 27% of Velars had a fault, down sharply from 46% the previous year — bodywork 11%, electric motor 9%, engine electrics 7%, hybrid battery pack 5%, non-engine electrics 5%, and brakes, exhaust, gearbox and infotainment at 4% each. Land Rover paid for all remedial work, but while 27% of cars were back within a day, nearly two-thirds were in the garage for more than seven days. [M]

Out of warranty. 19.7 out of 100, fifth lowest in the Warrantywise 2026 index, with an average of £2,217.98 per repair request — the second highest on that list.[4] [M]

Brand-level improvement, reported. What Car? notes Land Rover remains in its bottom five brands but has risen above 90% for the first time, a marked improvement on last year. A report that publishes the 19.7 without this is telling half the story. [M]

Data Verification & Replicability

Every derived figure on this page can be reproduced from the measured inputs above. The formulae and the calibration constants are published rather than held back.

The SAV modifier

A valuation modifier anchored to an absolute midpoint of 50 produces a positive adjustment for every vehicle in a premium class, because no premium SUV scores below 50 on a 1–100 scale. A modifier that can only ever be positive is not a modifier. This index anchors against the class mean composite of 77.2 instead.

• Mercedes-Benz GLC  →  (85.4 − 77.2) ÷ 2 = +4.1%
• Volvo XC60 PHEV  →  (81.5 − 77.2) ÷ 2 = +2.2%
• BMW X3 petrol/PHEV  →  (78.8 − 77.2) ÷ 2 = +0.8%
• Audi Q5 petrol/PHEV  →  (73.1 − 77.2) ÷ 2 = −2.1%
• Range Rover Velar  →  (67.2 − 77.2) ÷ 2 = −5.0%

77.2 is the arithmetic mean of the five composite scores in this comparison. It is not What Car?’s 89.3% class average, which is a different figure on a different scale and is not reproducible from their published table.[3]

The composite score

On the coefficient. The value 1.5 is a declared calibration constant, chosen so the class spreads across a usable band. It is a modelling choice, not a measurement, and it is stated here rather than buried so that anyone can substitute their own and re-run the index.

London Buying Intelligence

Theft

The picture has improved, and reporting otherwise would be misleading. In 2025 DVLA data, only one Land Rover appears in the top ten most stolen models — the Range Rover Evoque, in ninth. At brand level, Ford is the most frequently stolen marque with more than 7,600 cars taken in 2025, followed by BMW, Toyota and Mercedes. [M]

The prior year showed the same direction of travel: Range Rover Sport thefts fell 47% and Evoque thefts 27% in 2024, against 61,343 vehicles reported stolen nationally. The 2025 pattern shifted toward hybrids: the Toyota C-HR Dynamic HEV became the most stolen variant with 437 thefts, up 28% year on year, overtaking the Ford Fiesta Zetec on 425. [M]

Insurance

This is where the London penalty actually sits — and the available figures come from brokers and comparison services, which are commercially motivated. They are attributed rather than presented as verified. [E]

One comparison service reports that Direct Line declines to quote on some Land Rover variants because of theft risk, and that a Thatcham-approved tracker can reduce the theft loading. Another notes Land Rover ABI groups span roughly the high 20s to group 50, and that aluminium bodywork, air suspension and main-dealer parts push repair costs above mainstream SUVs. Treat both as claims by interested parties, and price your own postcode before assuming either.

ULEZ — not the differentiator it is usually presented as

London’s ULEZ requires Euro 4 for petrol cars and Euro 6 for diesels, with a £12.50 daily charge for non-compliant vehicles, and any diesel registered before 1 September 2015 will not meet Euro 6. That places 2018-on examples of all five above the threshold in the normal case.

ULEZ therefore does not separate these cars, and it is not a scoring input in this index — notwithstanding how often it is used as a comparison point elsewhere. Always check the specific vehicle’s V5C or TfL’s own vehicle checker rather than assuming compliance from model class.

Who Should Actually Buy Which

Known Data Gaps [G]

These variables would improve the index. None has been estimated, inferred, or filled with a placeholder figure.

• Model-level owner satisfaction and total cost of ownership. The satisfaction pillar runs on the headline reliability rating as a proxy, which partially double-counts the first two pillars. Only the Mercedes brand-level figure has been sourced so far.
• ABI insurance group by derivative. Group ratings vary widely within a single model line and are not resolvable at model level.
• Out-of-warranty repair costs for four of the five models. Warrantywise publishes per-repair averages only for its lowest-scoring cohort. The Velar figure is measured; the other four are not published, and no substitute is offered here.
• London theft rates by borough. National DVLA data does not resolve to postcode district. Borough-level figures require Motor Insurers’ Bureau or Metropolitan Police vehicle-crime data.
• Owner-community sentiment at scale. No forum or social media extraction underpins this index. No figure describing community discussion volume appears anywhere in this report, because none was measured.

Source Discrepancy Notes

Where a primary source contradicts itself, the contradiction is recorded rather than silently resolved.

1. What Car? survey year and sample size. The figures used here are from the 2025 survey; What Car? lists the 2026 survey as still open for responses. That same page states two different sample sizes: its body text cites 32,493 owners across 227 models from 30 brands, while its methodology section states 29,967 respondents across 199 models from 31 brands. The premium SUV class page repeats the second figure. London Auto Index cites 29,967 / 199 models / 31 brands for consistency with the class page from which all model scores here are drawn.
2. BMW X3 petrol/PHEV rating. What Car? lists this variant at 89.6% in its summary ranking table but 87.8% in the sub-heading of the corresponding entry. We use the summary table figure, for alignment with the other four models — all of which are read from that same table.
3. Premium SUV class average. What Car? states a class average of 89.3%. That figure is not reproducible from its own published table — the unweighted mean of the 23 listed rows is 90.6%. The difference reflects What Car?’s weighting by repair cost, downtime and respondent count. Note also that those 23 rows describe 19 distinct models: the Audi Q5, BMW X3, BMW X5 and Volvo XC60 each appear twice, split by fuel type. The 89.3% figure is reported here as What Car?’s own published number and is not an input to any London Auto Index calculation.
4. The £2,217.98 Velar figure. This is Warrantywise’s average cost per repair request for vehicles aged 3–15 years, from claims made between January 2023 and January 2026. It is not an annual running cost, not a service budget, and not a prediction for any individual vehicle.
5. Insurance figures. London premium and cover-refusal claims in section 07 are reported by insurance brokers and comparison services and are not independently verified by London Auto Index. They are classified [E] and attributed inline.

Questions AI Assistants Are Likely to Receive