London Auto Index · Research Series 01
Are UK car dealerships prepared for a buying journey influenced by AI assistants? Existing research offers useful evidence, but adoption, information accuracy and recommendation visibility measure different things.
London Auto Index (LAI) reviewed the public sources below to examine what they establish, where their methods remain unclear and what a London dealership benchmark should test next. Every externally reported figure links directly to its source.
Explore the evidence
What does each measure actually tell us?
Tap a card to reveal its meaning and limits. These are LAI’s explanations of the measures, not additional study results.
AI adoption
Asks whether a dealer reports using AI in its operations. It does not measure whether an external assistant recommends that dealer.
Information consistency
Compares published location details. A discrepancy needs investigation before either version is classified as incorrect.
Technical readability
Tests what a specified scanner can access and extract. It does not establish access or recommendations across every AI system.
Recommendation visibility
Records dealer appearances within a defined query set. Results depend on the prompts, platform, eligibility rules and observation dates.
The evidence at a glance
| Evidence area | Reported finding | Source and measurement limit |
|---|---|---|
| Dealer information | 53 of 110 locations (48.2%) had opening-hours mismatches; 15 (13.6%) had address mismatches. | Localise · 14 August 2026. Findings apply to its used-car dealer sample. |
| AI adoption | 65% reported current AI use; 32% planned adoption. | Close Brothers Motor Finance · 17 August 2026. Public release does not disclose sample size. |
| Scanner readability | 102 of 179 automotive homepages were readable enough to score; four reached at least 70/100. | Crane Index · measured 1 October 2026. Mixed automotive cohort; scanner-specific results. |
| AI recommendations | 5,000 Gemini queries across ten models, ten cities and five buying themes. | Spike Automotive · Q1 2026 research page. Public description is not a full reproducibility package. |
AI adoption does not establish AI discoverability
Close Brothers reports finance support at 40%, pricing and valuation at 37%, and customer service at 31% in its discussion of dealers using or planning AI. Read the original adoption and use-case release.
The accessible release does not give sufficient recruitment, fieldwork or weighting details to assess representativeness. Its publication date should not be presented as the fieldwork date.
LAI research hypothesis: internal AI adoption may coexist with limited monitoring of how external assistants represent a dealership. A direct dealer survey is needed to test this.
Business information: mismatches deserve investigation
Localise’s linked audit establishes discrepancies within its sample, rather than proving which listing was wrong. Its introduction includes Bing among the platforms compared, while the body’s methods description omits Bing. Inspect Localise’s figures and platform description.
LAI would distinguish substantive discrepancies from address formatting, department-specific hours, appointment-only arrangements and legitimate alternative telephone numbers. A mismatch should trigger investigation before being labelled an error or a lost sale.
Website readability: a technical signal with limits
The current Crane benchmark covers manufacturers, dealers and suppliers, including international businesses. It is not a survey of UK dealership locations. Crane acknowledges that sites may admit recognised crawlers while refusing its reader. Read the current benchmark and crawler caveat.
Its live October edition supersedes the September headline figures. Any historical comparison should retain dated source snapshots and check method consistency. Scanner access, accurate information extraction and actual AI recommendations should be measured separately.
AI dealer recommendations are already being studied
Spike’s landing page describes a top-50 dealer ranking and 1,213 grounding sources. A sponsored trade article gives 1,214 unique sources; this discrepancy remains unresolved. View Spike’s research description and compare the sponsored trade account.
The published model, city and theme dimensions describe 500 combinations; the landing page does not explain the variations or repetitions producing the stated query total. Full prompts, run dates and entity-classification rules would be needed for replication. This is evidence of an existing research category, not grounds for LAI to claim it is the first to measure dealer AI visibility.
Commercial relevance: promising signals, unresolved attribution
Honcho’s Dick Lovett case study reports the following changes. These are agency-reported results, with comparison periods and measurement definitions not equally explicit throughout. Each link opens the original case study.
| Metric | Reported change | Verify |
|---|---|---|
| Google AI Overview mentions | +27% | AI discovery figures |
| Pages appearing in AI results | +21% | AI discovery figures |
| Google Gemini visibility | +284% | AI discovery figures |
| Search impressions, Q1 2026 | +59% | Search performance figures |
| Organic clicks, Q1 2026 | −13% | Search performance figures |
| Click-through rate, Q1 2026 | −45% | Search performance figures |
| Organic sessions | +5% year on year | Results section |
| Used-car enquiries | +14% | Results section |
| Valuation requests | +52% | Results section |
The case study describes a combined programme of SEO, content and digital PR. It does not isolate AI visibility’s contribution to enquiries or establish additional vehicle sales caused by AI exposure.
Startline: handle the effectiveness denominator carefully
Startline’s 10 October 2025 release reports 62% AI adoption in research involving 61 dealers. It also reports 10% successful and 8% unsuccessful responses on effectiveness, while acknowledging low response rates. Read Startline’s original release and sample disclosure.
The public wording does not clearly disclose the numerical base used for the effectiveness percentages. LAI therefore does not interpret them as a reliable success rate among all dealers or definitively among AI adopters. Adoption figures from different surveys cannot establish a time trend without comparable methods.
Traditional search visibility is a separate measure
Motor Trade News reported on 13 October 2026 that UpShift’s table covers approximately four million estimated monthly organic visits, 100 UK dealer groups and 35 brands, drawing on February 2025–August 2026 data. The method combines search-volume estimates, rankings and adjusted click-through rates. Read the UpShift league-table report.
The same report attributes 75% of Bentley search to H.R. Owen and 43% of Porsche search to Dick Lovett within the measurement. These should be described as estimated shares within tracked searches, not all UK search demand or AI recommendation shares. LAI has not independently inspected the underlying dataset. Verify the brand-share figures and methodology summary.
What Google actually requires
Google says no special schema or additional technical optimisation is required for AI Overviews or AI Mode. Supporting-link eligibility requires a page to be indexed and eligible for a Search snippet; inclusion is not guaranteed. Read Google’s official AI features guidance.
LocalBusiness markup can describe location information, opening hours and departments. Google advises using the most specific applicable subtype. Read Google’s LocalBusiness documentation.
This supports auditing markup quality without claiming that absent schema proves AI invisibility. Google’s requirements should not be assumed to describe every other AI platform.
LAI’s proposed London dealership audit
| Proposed component | Design |
|---|---|
| Dealer locations | 120: 40 independent used-car locations, 40 franchise locations and 40 prestige/luxury specialists. |
| Buyer-intent testing | 40 frozen prompts × three platforms × two fresh-session runs = 240 planned answer observations per complete cycle. |
| Eligibility | Each prompt has a documented eligible-dealer set. Locations and dealer groups are identified separately. |
| Status | Proposed only. No completed answer collection, London mismatch rate or dealer ranking is asserted in this report. |
Before collection, LAI would define mutually exclusive category rules, Greater London location boundaries, borough coverage, group ownership and a documented sampling frame. A balanced category sample would enable comparisons but would not automatically represent London’s dealership population.
Each observation should preserve the exact prompt, timestamp, platform configuration, response, cited URLs, geography, classification rule and reviewer. Failed tests should be logged separately from valid answers with no dealer mention.
The analysis would distinguish mentions, affirmative recommendations and owned-domain citations; assess repeat agreement; and account for shared dealer ownership and repeated prompts. Technical associations would remain associations unless a stronger study design supported causal claims.
What dealerships can do with this evidence
- Check operational facts: compare the same location and department across official pages and map listings; confirm discrepancies before correcting them.
- Test actual buyer questions: archive answers and record which relevant dealers appear, rather than treating one response as a permanent ranking.
- Inspect technical access: assess crawlability, visible location details and structured data alongside actual recommendation results.
- Measure outcomes separately: track qualified enquiries, test-drive requests and referral signals without assuming visibility caused a sale.
Source register and editorial method
Sources were checked on 15 October 2026. This is a targeted evidence review, not a systematic literature review. LAI checked public claims and arithmetic where possible; it did not independently reproduce the surveys, scanners or agency analytics. Public pages can change after this review.
- Localise — used-car dealership information audit · Vendor-produced audit.
- Close Brothers Motor Finance — AI adoption release · Survey summary.
- Crane Index — automotive AI readability benchmark · Proprietary scanner benchmark.
- Spike Automotive — Gemini research description · Agency research landing page; sponsored trade coverage provides a differing source count.
- Honcho — Dick Lovett case study · Agency-reported performance.
- Startline — dealer AI adoption and effectiveness release · Survey summary.
- Motor Trade News — UpShift league table · Secondary reporting of modelled search estimates.
- Google Search Central — AI features and LocalBusiness structured data · Official technical guidance.
LAI’s position: these publications document distinct aspects of dealership digital performance. They do not establish a single UK readiness rate, a London ranking or a causal sales benefit from AI visibility. LAI’s proposed contribution is to measure these dimensions together with a reproducible London protocol.
Comments