Admiral Business is barely visible in the AI surfaces where SMEs are now researching insurance.

Admiral has credible presence in just 5 of 19 vertical-city cells. Designers in Cardiff is the strongest cell, at rank 1 with visibility 92. Outside South Wales, Admiral barely registers: ten cells return Admiral nowhere in the answer, and four more name it at rank 7–10, where it is cited but not chosen.

Admiral's position in each vertical-city cell, colour-coded by quality of presence. Strong means Admiral hit rank ≤ 3 with at least one provider; Mid means rank 4–6; Weak means Admiral was named only at rank 7–10 (functionally invisible in the answer). Absent means no LLM named Admiral at all. Click any cell to drill into the top brands, the source domains that drove those rankings, and Admiral's specific citation evidence in that cell.

Each of the 20 completed prompts, with the citation totals each one generated and Admiral's status in that specific query. Click any card to see the source domains that drove citations and the top brands the LLMs named in their answers.

The four major LLMs behave very differently. Each ranks regional specialists differently, cites different sources, and has a different relationship with Admiral. Treating them as one audience hides the real story.

Beyond who LLMs cite, the data captures how they cite. Across 1,706 ranking justifications, 94% are positively framed, so absolute sentiment is not the signal. The signal is the shape: which reason categories dominate (expertise, reviews, pricing) and which brand earns clinical neutrality versus marketing enthusiasm.

Three findings the data supports

What LLMs literally say about Admiral

These are verbatim reason strings from the citation_reasons table, attributed to Admiral Business by the named provider. Sample is small (3 positive, 1 neutral attributable quotes); interpret accordingly.

Sentiment + rank, top 12 brands

Mean sentiment compound score across all completed-job mentions, joined to ranking observations. Sorted by mention volume. Brands with fewer than 5 mentions excluded for stability. Admiral shown for reference despite small sample.

These are the source domains the four LLMs cited most heavily when answering Admiral-relevant prompts. Weighted influence accounts for both citation frequency and citation position (position 1 counts double position 5). Domains marked insurer are competitor first-party properties; comparison are aggregator/review sites; trade-body are NICEIC/FMB-class authorities.

Domain type share of weighted influence

Top 25 source domains

The source URLs the LLMs read when constructing answers that ended up naming Admiral. Important framing: these are not direct citations of admiralbusiness.com — they are the source pages the model consulted while building a ranking that happened to contain Admiral. Of the 253 source rows traced back to an Admiral ranking, only 11 are admiralbusiness.com self-citations. The rest are competitor pages, comparison aggregators, and review sites that mentioned Admiral in passing (often as a footnote in a "best UK insurers" list).

The strategic read: simplybusiness.co.uk, axa.co.uk, hiscox.co.uk and markeluk.com each surface Admiral more often than admiralbusiness.com itself does in the LLM source pool. Admiral's on-site signal is being out-performed by incidental mentions on competitor properties — a fixable structural weakness.

Three tiers emerge from the data. National giants (5 cities, 4 verticals) — Simply Business, Hiscox, AXA, Direct Line for Business, Markel Direct — collectively own ~22% of all ranked positions. Specialist nationals (Rhino, Tradesman Saver, PolicyBee, Kingsbridge) win specific verticals — tradespeople, designers, contractors — across geographies. Regional specialists win one city across multiple verticals (see § 07).

For every ranking observed, the pipeline extracted the model's stated reason for citing that brand. Below: the categories of reason the LLMs use, vs. the categories of content gap the pipeline identified on admiralbusiness.com.

Top 10 highest-priority gaps

Next steps

(1) Commission the three Phase 1 flagship microsites in § 08 and brief the Bloxx component pipeline for Phase 0 hero proof-of-concepts. (2) Expand the audit to 300+ prompts across all four buyer-journey stages (problem-aware, solution-aware, comparison, ready-to-buy) and the extended regional matrix, adding the Couriers vertical. (3) Set up weekly citation share drift monitoring once the Phase 1 microsites go live so we can measure the lift in real time before committing Phase 2 capital. (4) Add Scotland, Northern Ireland, the North East and Bristol to the regional matrix in the full run.