The UK insurance AI Visibility Index
What this Index measures
The AI Visibility Index measures how often and how favourably each brand appears in the answers given by AI assistants to common insurance questions. It is not a measure of price, cover, claims service, financial strength or value for money, and it is not a recommendation to buy any product. A high score means a brand is well represented in AI answers. It does not mean the insurance is better.
Independence and conflicts
- No brand pays to be listed, removed or re-ranked. Positions are measured from real AI responses.
- Marrow sells infrastructure to insurers and may have a commercial relationship with brands appearing in the Index.
- Marrow is not an insurer or a broker, and is not authorised by the Financial Conduct Authority.
Informational only. Not financial advice, not a personal recommendation, and not a comparison of price or cover. Always check terms directly with the insurer.
Top 11 of 11 brands · Updated 2026-08-10
- #151475999190
Admiral
IntermediaryCarrier details incompleteUnderwritten by a panel of 3 carriers: Admiral Insurance Company Limited, Admiral Insurance (Gibraltar) Limited, Great Lakes Insurance UK Limited. Insurer details
Score 51 · Vis 47 · Sen 59 · Acc 99 · Rec 19
- #248446191145
Aviva
InsurerUnderwritten by Aviva Insurance Limited. Insurer details
Score 48 · Vis 44 · Sen 61 · Acc 91 · Rec 14
- #345346097121
Direct Line
InsurerUnderwritten by U K Insurance Limited. Insurer details
Score 45 · Vis 34 · Sen 60 · Acc 97 · Rec 12
- #44433619791
LV=
InsurerUnderwritten by Liverpool Victoria Insurance Company Limited. Insurer details
Score 44 · Vis 33 · Sen 61 · Acc 97 · Rec 9
- #53917669551
NFU Mutual
InsurerUnderwritten by The National Farmers Union Mutual Insurance Society Limited. Insurer details
Score 39 · Vis 17 · Sen 66 · Acc 95 · Rec 5
- #63820569931
Hastings Direct
IntermediaryCarrier details incompleteUnderwritten by Advantage Insurance Company Limited. Insurer details
Score 38 · Vis 20 · Sen 56 · Acc 99 · Rec 3
- #73811639875
Saga
IntermediaryUnderwritten by Ageas Insurance Limited. Insurer details
Score 38 · Vis 11 · Sen 63 · Acc 98 · Rec 7
- #83716559930
Churchill
InsurerUnderwritten by U K Insurance Limited. Insurer details
Score 37 · Vis 16 · Sen 55 · Acc 99 · Rec 3
- #93513539720
AXA
InsurerUnderwritten by AXA Insurance UK plc. Insurer details
Score 35 · Vis 13 · Sen 53 · Acc 97 · Rec 2
- #10325559320
Tesco Bank
IntermediaryUnderwritten by Tesco Underwriting Limited. Insurer details
Score 32 · Vis 5 · Sen 55 · Acc 93 · Rec 2
- #11263527500
Co-op
IntermediaryCarrier details incompleteLimited dataUnderwritten by Underwriting carrier not established for this line. Insurer details
Score 26 · Vis 3 · Sen 52 · Acc 75 · Rec 0
Showing the top 20 brands by AI Visibility Score. The Index ranks consumer brands, not insurers. The company each contract of insurance is made with is named on every row.
AI Visibility Scores are derived from live AI assistant outputs, which vary by prompt, session and model version, so they are a directional signal of AI visibility rather than a guarantee of how any single query will respond. The Index is editorial and unsponsored. This is not financial advice; always confirm policy terms directly with the insurer.
Direct Line and Churchill are both underwritten by U K Insurance Limited and are ranked here as separate consumer brands, not as separate insurers.
AI visibility by insurance line
The same measurement, broken down for each line of cover.
How the AI Visibility Score is built
We measure how UK insurance brands show up when consumers ask AI assistants about insurance. Each quarter we run a fixed battery of prompts across ChatGPT, Claude and Gemini, mirroring the real questions shoppers type ("best car insurance for young drivers", "cheapest pet insurance for a puppy", etc.) and the follow-ups they ask next.
Every response is parsed for brand mentions. Each mention is classified on four dimensions: whether the brand appeared at all (visibility), whether the surrounding sentiment was positive, neutral or negative (sentiment), whether the claims the model made about the brand were factually correct against public FCA and provider data (accuracy), and whether the brand was presented as a top pick (recommendation).
Each dimension is normalised to a 0-100 scale and weighted into a single composite AI Visibility Score. The Index is refreshed quarterly. Every dimension describes the content of an AI answer. None of them measures price, cover, claims service, financial strength or value for money.
A brand is only given a position if it was measured on enough prompts in that line to place it fairly. Brands that appear too rarely are still listed with the scores we measured, but shown as unranked rather than slotted above brands measured several times as often. This is a deliberate change from earlier editions, which ranked every brand that produced any signal at all.
Each edition measures the models consumers are actually being served, so the specific model versions change from quarter to quarter as the assistants themselves are updated, and the depth of sampling behind each figure can change with them. That means quarter-on-quarter movement reflects what changed about the brand, what changed about the models, and how the measurement itself was run. Where a brand moves sharply, the model refresh is as likely an explanation as anything the brand did.
The Index ranks consumer brands, not insurers. Many of the brands listed are intermediaries rather than the company the insurance contract is made with, and some are underwritten by a panel of carriers whose composition varies by policy. The underwriting carrier for each brand is published alongside it.
What is an AI Visibility Score?+
A 0-100 measure of how a brand appears when consumers ask AI assistants about insurance. It combines visibility, sentiment, factual accuracy of the model's claims, and how often the model presents the brand as a top pick, per insurance line. A high score means the brand is well represented in AI answers. It does not mean the insurance is better.
What does the Index not measure?+
Price, cover, policy terms, claims service, complaint rates, financial strength and value for money. None of these are inputs. Nothing in the Index should be read as a recommendation to buy, or as a comparison of the insurance products themselves.
Which AI engines are measured?+
ChatGPT, Claude and Gemini, each on the model generation a consumer is served today rather than a version pinned to an earlier edition. The specific models are refreshed every quarter, which is why movement between editions reflects model change as well as brand change. Perplexity is on the roadmap once UK consumer usage of agentic search crosses our inclusion threshold; we add engines as the evidence supports it.
Which life insurance products does the Index measure?+
Pure protection only: level and decreasing term life, critical illness cover, and over-50s guaranteed acceptance plans. Investment-linked whole-of-life policies and endowments are excluded, because they are packaged investment products rather than pure protection and sit under a different regulatory framework from the one this page is published under. Life cover is medically underwritten and priced individually, so nothing on this page indicates what any policy would cost or pay out for you.
Why is a brand listed but unranked?+
Because too few prompts in that line produced a usable signal about it to justify giving it a position. A score built on one or two prompts is not comparable to one built on ten, and placing them side by side would imply a precision the measurement does not have. Unranked brands are still shown with the scores we measured. From Q3 2026 the Index applies this floor; earlier editions ranked every brand that produced any signal, so some positions are not directly comparable across that boundary.
How often are scores refreshed?+
Quarterly. LLM behaviour shifts with every model release, so fresh measurement matters more here than in traditional SEO.
Why is a brand ranked in a line it does not sell products in?+
Because the Index records which brands AI assistants name when asked about a line, not which brands offer products in it. Where a model names a specialist brand outside its actual market, that mention is still measured. Read a position as a fact about AI behaviour rather than as a statement about the brand's product range.
Can a brand pay to improve its position?+
No. Nothing on this page can be purchased. Scores are measured independently from public AI outputs, and Marrow's commercial relationships with any insurer have no bearing on how that insurer is scored.