Best AI visibility tool for an insurer, bank or wealth manager, and which entity gets named
The answer may name the institution, the branch or one adviser on a public register, so the tool has to count three different entities.
Published by AI Knows Us · Updated 20 October 2026
For a regulated financial institution the deciding feature is whether the tool can tell you which entity an assistant named: the brand, a branch or office, or an individual adviser, broker or agent. In this sector the three are often separate legal entities with separate register entries, and a tool that reports one brand score averages them into a number nobody can act on. Buy on what each vendor publishes about entity and location handling, which you can read on their own pages. Everything below was read on 20 October 2026.
One page for insurance, banking and wealth management, and why
These three were separate rows on our build list and they are one page here, because the answer is the same answer. In each one the buyer finds a named human or a named local office through an assistant, the credential that settles trust is an entry on a public register, and the institution's own brand is often not the thing the answer names. A page per sector would have split one query three ways and said the same thing in three places.
Where the sectors genuinely diverge is in which register and which intermediary, and that belongs in the prompt set rather than in a separate page. A fintech is a different purchase and has its own page, because a fintech usually sells a software shaped product and frequently operates under somebody else's licence.
What the engines actually said
In our audit of 1,260 readings across ChatGPT and Gemini in India, the United Kingdom and the United States, the question how does an insurance broker appear in AI recommendations named exactly one rival, Semrush, in one reading. The question how do financial advisers get named by AI assistants named Peec once. Our own name appeared in neither.
Compare that with software, where the same audit returned Profound in six readings, Semrush in six, Otterly in five and Peec in five on a single question. In regulated finance the engines barely name any tool at all. The category has published almost nothing for this buyer, so the recommendation slot is close to empty. That is a measured finding about a gap, not a claim about quality, and it is the most useful thing on this page if you are choosing when to start.
The entity problem, concretely
Run one prompt three ways and you will see it. Ask for the best insurance broker in a named town, then ask about your firm by name, then ask about one of your advisers by name. In our city level readings the winning sources were frequently neither the national brand nor the local firm: in several American and British cities a national publisher or a city named domain filled the slot instead. We published those city readings individually, for example the Birmingham reading, where a national newspaper filled the slot.
So the measurement you need is a matrix, not a score: entity down one axis, location across the other. That shape is what sets your tier at every vendor, and it is why the limits below matter more here than the headline figure does.
What each vendor publishes, with the limits that decide this vertical
Fetched with a macOS Safari agent and again with a Chrome on Windows agent. We record limits and leave prices on the vendor's own dated page.
| Vendor | Location limit published | Persona or entity limit published | Method or failure published |
|---|---|---|---|
| Scrunch | Yes, a country count on its entry tier, custom above it | Yes, a persona count, and the entry tier includes one | Responses collected a month is published; no sample size or negative result in the document read |
| Peec | Yes, country and language counts per tier, with multi country named on a middle tier | Competitor count published per tier | Publishes the arithmetic behind its headline unit; no negative result in the document read |
| Otterly | Yes, markets named in the tier table | Not in the document read | Not in the document read |
| Profound | Yes on the trial, one region and one language, custom above it | Not in the document read | Not in the document read |
| Semrush | Not on the AI visibility feature page; a local SEO product is named separately in its navigation | Not in the document read | Not in the document read |
| Ahrefs | Not on the Brand Radar page | Not in the document read | Not in the document read |
| AI Knows Us | Yes, and we publish the city by city readings themselves rather than a capability claim | We publish the readings per entity type | Yes. 1,260 readings, 4,947 domains, and our own 6 citations, zero in the United States and zero on Gemini |
Read that table as a disclosure ranking and nothing else. We are first on published method and published failure. Rivals are ahead of us elsewhere and it matters for a regulated buyer: Otterly publishes a monthly and an annual figure for each self-serve tier and prices each extra engine separately, Peec publishes a changelog and a trust page, Profound and Scrunch name a recognised security certification, and our own domain carries no trust centre, status page or changelog as of 20 October 2026. A bank's vendor assessment will notice that before it notices our reading counts.
If you are checking these pages yourself, otterly.ai/pricing/ answered 200 to a macOS Safari agent and 403 to a Chrome on Windows agent on 20 October 2026, with the agent string as the only variable, so change agent before recording a refusal.
How to size the purchase
- Count your entities first. One brand, plus the offices or branches a customer would search for, plus the advisers who hold their own register entry and have their own page.
- Then count locations. Entities multiplied by locations multiplied by about four questions is your realistic prompt count, and it is usually larger than a marketing team expects.
- Then check the country limit. An entry tier that includes one country is fine for a single market firm and useless for a group with offices in three.
- Then check who gets the export. In this sector compliance will want the answer text, so the export tier is part of the price rather than an upgrade.
Questions insurers, banks and wealth managers ask about this
The assistant names one of our advisers and not the firm. Is that a problem?
It is a result, and usually a good one, because a named individual with a register entry is exactly what this buyer wanted. Measure it deliberately rather than treating it as noise, and make sure the adviser's page names the firm so the answer can carry both.
Should we publish our register number on every page?
Publish it where a reader is deciding, in text rather than in an image, and name the register. A number somebody else can confirm is the single most repeatable fact a regulated firm owns, and our local service readings found the same pattern across markets.
Do comparison and aggregator sites decide this category?
They appear, and they are not the whole picture. Across our second audit, vendor owned pages carried 82.0% of all citations and review directories carried 0.7%. The aggregator is loud in this sector precisely because so few firms publish the facts themselves.
Our products differ by state or by region. How many prompt sets is that?
One per region you are authorised in, at minimum. The answer an assistant gives changes with the location in the prompt, so a single national reading will tell you about the largest market and nothing about the others.
Is there a page that covers what to publish, rather than what to buy?
Yes, per trade. The broker version is how an insurance broker gets recommended by AI, and the wealth side is how a wealth manager gets recommended by AI.
How often should a regulated firm re-read this?
Monthly, and after any product or rate change, because that is the moment an old answer becomes a wrong answer about a current product.
The first reading to take, before any vendor call
Pick your largest office. Ask an assistant for the best firm of your type in that town, three times, with no names in the prompt. Write down the firms named and the exact URL cited for each. If a national publisher or a directory filled every slot, your problem is that nothing you publish is specific to that town, and that is a writing problem a dashboard will describe but not solve. AI Knows Us takes that reading per office and per named adviser, and the readings behind our own published claims, including the empty ones, sit on our proof page.