London Auto Index / Research governance
Every useful answer should show what supports it, when it was checked and what the buyer still needs to confirm.
Version 1.0 · Effective 23 October 2026
Editorial owner: Hassan Mehmood · Scope: vehicle, journey, dealer, stock, AI discovery and fulfilment research.
1. Purpose and editorial responsibility
LAI researches how London buyers can connect their requirements with suitable vehicles, dealership routes and supporting evidence. Each research page must identify its scope, author, evidence dates, sources and material limitations. Coverage of a business does not establish a partnership or endorsement.
Author profiles must describe relevant experience accurately. Desk research must not be presented as a road test, vehicle inspection, luggage test, dealership visit or completed buyer transaction. Where specialist review is needed, the reviewer and their actual contribution must be identified; unreviewed work must not imply independent expert approval.
2. Evidence labels readers can interpret
| Label | Meaning | Boundary |
|---|---|---|
| Primary-source fact | A dated manufacturer, public authority or other originating document supports the precise statement. | State the applicable model year, variant, geography and conditions. A published specification is not an LAI physical measurement. |
| Dealer statement | The dealer publishes or directly supplies the information. Identify the channel and date. | Confirms what the dealer states; does not independently prove service quality, inspection work or availability. |
| LAI observation | LAI directly recorded a listing, interface response, visit or other event using a described method. | Applies to that observation and research window; does not establish an unseen cause. |
| LAI calculation or interpretation | An estimate, comparison or conclusion derived from identified inputs. | Show assumptions, formula where relevant, trade-offs and uncertainty. |
| Unverified / awaiting confirmation | Evidence is missing, conflicting, inaccessible or insufficient for the claim. | Exclude the claim from confirmed eligibility and highlight the check required. |
Evidence type and freshness are separate. A dealer statement can be recent but uncorroborated. An original manufacturer document can be authoritative but apply to a different generation.
3. Minimum evidence record
For each material factual claim, the working record must capture: claim ID; exact statement; relevant dealer, vehicle, stock or buyer-intent IDs; source URL or document reference; a supporting extract or locator; publisher; evidence type; observation date and time where relevant; applicable variant and location; limitations; researcher; next review date; and the pages or recommendations using it.
Public pages must expose enough of that record for readers to check the important claims: source links close to the claim or in a keyed evidence table, dates, scope and unresolved questions. Private correspondence and personal buyer details must be protected; publish a permitted summary rather than private messages.
A broken or inaccessible source must trigger review. Conflicting evidence must be disclosed and resolved by scope, date and source authority, rather than selecting the most convenient figure.
4. Freshness and review rules
The following are editorial review thresholds, not a claim that automated monitoring is already operating.
| Information | Required treatment | When the threshold is exceeded |
|---|---|---|
| Stock and advertised asking prices | Timestamp every observation. Recheck within 24 hours before using a listing in an active shortlist; obtain current dealer confirmation before describing availability as confirmed. | Label as a historical listing snapshot. Never call it live or currently available without a current check. |
| Offers and finance examples | Record expiry, eligibility and full stated assumptions. Check on publication and before an active introduction. | Withdraw expired offers from current recommendations; retain only as labelled historical evidence. |
| Branch details, opening hours and published dealer programmes | Review at least every 30 days and before a specific visit or introduction. | Mark review overdue and direct readers to the current official route. |
| Vehicle specifications | Review at least every 90 days, and sooner on a model-year or variant change. | Mark review overdue; preserve the original source date and generation. |
| Charges, policy and regulatory information | Check the responsible authority on publication, at least every 30 days and on announced changes. Separate proposals from rules in force. | Flag overdue review and link to the authority. Confirm applicability before an individual decision. |
| AI discovery results | Always publish the actual test window and conditions. | Keep as historical findings. A fresh measurement requires a new documented run. |
Changing a page date does not refresh its evidence. A removed listing is a recorded removal; a completed sale requires separate confirmation.
5. Research and comparison methods
- Vehicle comparisons: align generation, trim, powertrain, body style and price basis. Explain differences such as mirrors included versus excluded, boot measurement methods and test-cycle versus observed range.
- Cost scenarios: show mileage, energy price, consumption and other inputs. Label estimates as scenarios and separate them from quotes or observed ownership costs.
- Stock studies: declare dealer sample, selection rules, collection times, duplicate handling, missing observations and sample size. Asking prices are not transaction prices.
- Buyer research: disclose recruitment, sample size, questions and whether behaviour was observed or self-reported. Research prompts and keyword-tool data are not proof of individual purchase intent.
- Physical tests: identify the vehicle and conditions, instruments, repeat procedure and supporting photographs where appropriate.
- Scoring: publish criteria, denominators, weights and exclusions before interpreting results. Missing data must not silently become a zero or a positive result.
6. AI discovery measurement
Retain the exact prompt, platform, visible model where available, timestamp, session conditions, answer evidence and assessment rules. Separate branded recognition from unbranded discovery, a mention from a recommendation, and an answer-body link from an exhaustively inspected citation set.
Web-search research must not be described as a genuine response from another assistant. Inaccessible platforms and failed runs must be disclosed. Small panels describe those observations only. A before/after change does not by itself prove that LAI publishing caused it.
The AI Visibility Methodology sets out the specialised index protocol and its validation status. Its provisional scores remain provisional until validated; this standard does not convert them into final rankings.
7. Buyer matching and fulfilment
A shortlist must identify the buyer’s stated requirements, eligibility criteria, supporting evidence, alternatives, trade-offs and unresolved checks. Separate vehicle suitability, dealer capability, advertised stock and directly confirmed availability.
LAI must not claim an appointment, dealer response, lead conversion or sale without a corresponding record. Obtain buyer consent before sharing identifiable details with a dealer, explain what will be shared and disclose any introduction fee. Aggregate research must avoid exposing individual buyers.
8. Commercial independence and disclosures
Businesses cannot buy a higher research score, editorial recommendation or favourable correction. The same published criteria apply to paying and non-paying businesses within a comparable study.
Any relevant paid diagnostic, consulting relationship, sponsorship, affiliate arrangement or introduction fee must be disclosed on the affected page or at the relevant action. Sponsored material must be visibly identified. Research coverage must not imply a commercial relationship where none has been established.
Payments fund specified services; they do not substitute for evidence. Dealer-supplied corrections are assessed on their supporting information.
9. Verification language and images
Verification applies to a named claim, method, date and scope. For example, “published warranty terms reviewed on [date]” means those terms were checked; it does not prove that every car has been inspected or that future claims will be accepted. A general dealer-quality badge requires a separately published assessment protocol and supporting checks.
Contextual stock photography and generated illustrations must not be presented as a photographed dealership, inspected vehicle or confirmed stock item. Attribute images where required and label illustrations clearly. AI assistance may support drafting and analysis; the named editor remains responsible for checking material claims.
10. Corrections and disputes
Readers and businesses can report an error. Include the page URL, exact statement, proposed correction and supporting evidence. The existing correction route states receipt within two working days and a decision within ten working days; these are service targets, not evidence of past performance.
Materially harmful or decision-critical disputed details should be flagged or withdrawn while checked. Record the decision and evidence. A material correction must show the date, what changed and whether a conclusion or score was affected. Minor spelling fixes need not be presented as new research. Apply accepted changes to dependent pages and records.
11. Publication checklist
- Define the buyer question, research scope and applicable entities.
- Check every material claim against its evidence and correct variant.
- Separate facts, dealer statements, observations, calculations and unknowns.
- Show source dates, review status, methods and sample sizes.
- Explain limitations and disclose relevant commercial relationships.
- Check source links, calculations, image descriptions and buyer action routes.
- Identify the author and any actual reviewer; make the correction route accessible.
- Record publication version, next review date and dependent outputs.
12. Adoption and change log
23 October 2026 — Version 1.0: introduced the cross-layer evidence taxonomy, minimum claim record, review thresholds, publication checklist and correction rules. Legacy pages require individual review; no site-wide compliance percentage is asserted.
Next policy review due: 23 January 2027, or earlier if the operating method changes materially.