Money questions are where AI models are at their most careful — and their most directory-shaped. Ask “best mortgage broker for self-employed, two years' accounts” and the answer leans on comparison giants and money-advice publishers, with brokers themselves barely present. That caution isn't bad news for you. It's the reason the door is open: a model that hesitates to name a broker it half-understands will happily name one whose regulatory standing and specialism it can verify in text.
Why brokers lose the answer today
Three patterns dominate the broker sites we scan. First, compliance-flattened copy: pages so carefully generic (“whole of market advice tailored to you”) that a machine can't tell what you're actually good at. Second, credentials as decoration — FCA numbers in a footer image, trade-body logos as badges — invisible to a crawler, when they're precisely the trust evidence a model wants before recommending a financial firm. Third, no specialism pages: the broker who does brilliant work for contractors, adverse credit or bridging has one Services page that mentions everything and therefore wins nothing.
What earns the citation in finance
- A page per specialism, answering first. “Self-employed with one year's accounts? Here's what's realistic, what lenders look for, and what we've placed recently” — that page can beat a comparison giant on its specific question, because the giant's page is generic by design.
- FCA registration as machine-readable fact. Your FCA number, permissions and trading names in structured data and page text, matching the register exactly. For money queries, models weigh verifiable regulation heavily.
- Named humans. Advisers with qualifications (CeMAP etc.) as structured Person data. “Who is advising me?” is a trust question machines ask on buyers' behalf.
- Third-party corroboration. Your entries on the money-advice directories, review platforms and trade bodies, linked from your own structured data so the evidence connects to you.
- Plain numbers where compliance allows. Typical fees, response times, lender panel size. Every concrete fact is quotable; every vagueness is a reason to cite someone else.
Measure it before and after
The free 60-second check shows whether AI can read your site at all — crawler access, structured data, entity clarity. The £19 citation audit asks all five major models twelve real borrower questions for your specialisms and patch, and names who's winning them. Because citations follow fixes by two to four months in this sector like every other, the honest play is the three-run pack: baseline now, fix, then prove the movement — a before/after your compliance file will actually like.