Diagnostic · Insurance brokers & agencies

What AI visibility looks like for insurance brokers.

A business asks an AI which broker to call before it ever picks up the phone — and the five engines hand back five different shortlists. No broker owns the answer; even the global names are surfaced inconsistently. The same diagnostic instrument finds different gaps here.

Answerability measures whether AI systems can find, trust, and cite your company — and shows what to change when they don't. This page is what that looks like for commercial insurance brokers and agencies. It measures observed AI surfacing behavior — not broker quality, suitability, or recommendation quality.

A risk manager or a founder used to start with a referral or an incumbent relationship. Increasingly the first pass is a prompt: "which broker should a venture-backed startup use for cyber and E&O coverage?" — and the model returns names. The question for a broker is no longer only whether your reputation is strong. It is whether the system doing the first cut surfaces you — and the data says that depends heavily on which engine the buyer happened to ask.

01What the engines actually do

We asked the five major engines which commercial insurance brokers to consider across six buyer situations, three runs each. The answer barely held still.

5 engines · web-grounded · 3 runsCaptured 2026-05-28
Which commercial insurance brokers are commonly recommended for [a mid-market manufacturer / a startup buying cyber & E&O / a multi-location business / …]?

Named in total

72 brokers across the six questions — from the global names to dozens of regional and specialty shops.

Most surfaced

Hub International, Marsh, Aon, Gallagher — but each swings hard by engine (Marsh: every ChatGPT run; only about half of Gemini's and Perplexity's).

Inter-engine overlap

0.38 (Jaccard); no prompt produced the same top broker across all five engines. "Molten" — the answer has not set.

What decided it
The engines have personalities: ChatGPT returns the global brokers on almost every prompt, while Perplexity ranks those low and surfaces regional and specialty firms — drawn from agency directories and local "best brokers" lists — that no other engine names.
Summary of the commercial-insurance-brokers edition of the Answerability Index. Observed surfacing across five web-grounded engines, three runs each, 2026-05-28 — not a ranking, endorsement, advice, or suitability judgment. Carriers were excluded; brokers only.

Two facts in that capture drive everything below. First, the category is molten — there is no consensus broker, so being surfaced is contestable rather than locked. Second, which broker you get depends on which engine you ask: a buyer on ChatGPT and a buyer on Perplexity are handed different firms for the same need — so a broker can be "visible" on one engine and absent from the rest, which is most of the buyers.

02Why brokers are uniquely exposed

Most categories just want to be found. A broker has to be trusted first — insurance is regulated and high-stakes, so the engines lean on independent corroboration over a firm's own claims — and then surfaced for the right line and segment. Three things make that hard.

Authority is offline and relationship-bound. A broker's reputation lives in renewals, referrals, and carrier relationships the engines cannot see. What they can see — earned media, directories, rankings — is exactly what most brokers have never built deliberately.

The answer splits by line and segment. Cyber for a startup, D&O for a board, contractor liability, multi-location property — each is a different question with a different surfaced set. One "commercial insurance" page competes weakly in all of them at once.

Your entity is ambiguous. Insurance is a roll-up industry: acquired agencies, regional brands, and parent groups blur together, so the engines struggle to resolve which firm is which — and the credit for a strong local reputation doesn't attach.

The result is the molten field above: real brokers, surfaced by one engine and invisible to the next.

In a molten category, the engines don't reward the best broker. They reward the one each engine can resolve and corroborate — and they don't agree on who that is.

03Where the answer is decided — the three pillars for a broker

Answerability is the composite a firm earns across three independent pillars, and it is capped by the weakest one. For most brokers the constraint sits in Trust and entity clarity — the reputation is real; the engines just can't corroborate or resolve it.

PillarWhere it lives for a brokerThe common gap
ContentLine-of-business and industry pages (cyber, D&O, construction, healthcare), claims and process pagesOne generic "commercial insurance" page instead of segment-specific authority; the buyer's actual question is never answered answer-shaped
RetrievalAgency directories, the firm's own site, carrier and association listingsThe directory profile is read instead of the firm; team and specialty pages thin or gated
TrustEarned media, rankings ("best brokers" lists), associations, a resolvable entity across acquired brandsThe binding constraint: thin third-party corroboration, no clean entity, authority concentrated offline or on pages engines don't reach
A typical broker: adequate content and retrieval, weak trust TYPICAL COMMERCIAL BROKER · ILLUSTRATIVE Content 62 Retrieval 72 Trust 24 binding constraint
Illustrative. The weakest pillar caps the composite. For a broker whose authority is offline and whose entity is split across acquired brands, Trust typically binds — more service pages will not move it.

04What the Diagnostic gives you

The AI Answerability Diagnostic measures your firm on the questions your prospects actually ask — line by line, segment by segment. It is a written intelligence report, not a dashboard.

See what the report looks like — a sample, section by section →

05What we usually find

The patterns recur. You will likely recognize at least one:

See where the engines route your buyers

The Diagnostic runs your real, line-specific buyer questions across all five engines and scores every cited source — so you can see where you're surfaced, where a ranking or competitor is named instead, and what moves it. The Diagnostic then continues as monthly Visibility Intelligence; the Sprint is the done-for-you build.

Evidence standard The capture summarized here is real: the commercial-insurance-brokers edition of the Answerability Index, five web-grounded engines, three runs each, 2026-05-28, sources recorded verbatim; carriers excluded, brokers only, names canonicalized. It records observed surfacing in a bounded sample — not a ranking, endorsement, broker suitability, or insurance advice, and not a longitudinal measurement, which is why the Diagnostic includes monthly Visibility Intelligence. How the audit works is set out in our methodology.

Diagnostic · Insurance brokers & agencies · hello@answerability.ai · Confidential under MNDA