AI assistants recommend mortgage brokers they can verify. When someone asks ChatGPT, Gemini or Perplexity for a broker in Parramatta or a first home buyer specialist in Brisbane, the assistant does not guess. It pulls from sources it already trusts: your website, your Google Business Profile, your aggregator listing, industry directories, review sites and articles that mention you. Brokers who get named are the ones whose details match everywhere, who are clearly local, who say specifically what they do and for whom, and who are cited by third parties. Everything below is about becoming that broker, one factor at a time, with work you can do this month.
Why broker questions are now asked of AI assistants
The questions people used to type into Google are now typed into a chat box. "Best mortgage broker in Chatswood." "Do I need a broker for my first home?" "Who can help me refinance if I am self-employed?" Google now answers many of these with an AI Overview above the first link. ChatGPT and Perplexity answer with a short list of names and a reason for each.
A person who asks an assistant for a recommendation and gets your name is a warm, exclusive enquiry. If the assistant names three competitors and not you, that enquiry went elsewhere before you knew it existed. For a broker who buys leads, this is the clearest path to owning the top of the funnel instead of renting it. The broader approach is on our financial services marketing page.
How ChatGPT, Gemini and Perplexity pick sources
Each engine works a bit differently, but the pattern is the same: retrieve, check, summarise.
- ChatGPT uses its own crawler and a search index for current questions. For local recommendations it leans on business listings, review sites and pages that state a location and a service plainly.
- Gemini and Google AI Overviews draw on Google's index, Google Business Profile data and Google reviews. If you already rank for local broker terms and your profile is complete, you are closer than you think.
- Perplexity searches the live web for every answer and cites its sources inline. It favours pages that answer the exact question in the first few lines and that look authoritative: named people, credentials, dates, specifics.
The common thread is corroboration. One source saying you are a broker in Hornsby is a claim. Five sources agreeing on your name, suburb, phone number and licence is a fact an assistant will repeat. We explain the differences from classic rankings in generative SEO vs traditional SEO.
The broker-specific trust signals
Assistants are cautious with financial topics. They want evidence you are a real, regulated operator before they put your name in an answer.
Credit licence or credit representative number
Put your Australian credit licence number, or your credit representative number and the licensee you operate under, in the footer of every page and on your about page. ASIC's credit licensee pages explain the two arrangements. It needs to be plain text, not an image, so a crawler can read it.
Your aggregator, named
State which aggregator you write through. It is one more verifiable fact that ties your business to a known entity, and your aggregator's own broker directory is often one of the sources an assistant checks.
MFAA or FBAA membership
Name your industry body on the site and make sure your listing in their "find a broker" directory is current and matches your website exactly. The MFAA and FBAA directories are exactly the kind of third-party corroboration these engines reward.
Google reviews with specifics
Detail beats volume. A review that says "Sarah got our first home loan approved in Penrith when the bank said no" contains a name, a loan type, a suburb and an outcome. That is quotable. Ask clients to mention what you helped with and where. Reply to every review, because replies are crawled too.
Named brokers, not a logo
Assistants recommend people more readily than brands. Every broker in the business needs a profile page: photo, full name, credit representative number, years in the industry, loan types they focus on, suburbs they cover, and a short plain-English bio. Link that page from the author line of any article they write.
A quotable page for each loan type and suburb
This is the biggest piece of work and the one that pays off longest. For every loan type and area you genuinely serve, you want one page an assistant can lift an answer from. Not a hundred thin pages. A handful of good ones. Start with the pairs that match your real settlements.
Each page should have:
- A direct answer in the first 120 words. Who you help, with what, where, and what makes you the right call. Write it the way you would say it to a client on the phone. The assistant needs a complete, liftable paragraph before it reaches any marketing copy.
- Specifics an assistant can repeat. Lenders on your panel, the suburbs you meet clients in, whether you do evening or video appointments.
- An FAQ section answering the five or six questions you hear most for that loan type. Short questions, short answers, no hedging.
- Structured data. Mark the business up as FinancialService with name, address, phone, area served and opening hours. Mark each broker up as Person with job title and the organisation they work for. Mark the FAQ up as FAQPage. Schema does not make a weak page strong, but it removes doubt about what the page is and who is behind it.
- A named author who is a real broker in the business, linked to their profile page.
Our ChatGPT optimisation service builds these pages for clients, but the structure is simple enough to brief a writer on yourself.
Let the crawlers in
None of this works if the assistants cannot read your site. It is the most common failure we find on broker websites, and it is usually accidental.
Check two things. First, your robots.txt. It should allow GPTBot and OAI-SearchBot (OpenAI's crawlers, documented at platform.openai.com/docs/bots), PerplexityBot (documented on Perplexity's crawler page) and Google-Extended (listed among Google's crawlers). Some website templates and SEO plugins block AI crawlers by default. If you see a Disallow line for any of these, remove it.
Second, your firewall or CDN. Services such as Cloudflare offer a one-click setting that blocks AI bots, and some hosts turn it on without asking. If AI bots are blocked, your pages do not exist as far as ChatGPT or Perplexity are concerned. Google AI Overviews still see you through the normal Googlebot, which is why some brokers appear in Google's AI answers and nowhere else.
Same name, address and phone, everywhere
Unglamorous, and for a local business it matters more than anything else on the list. Assistants cross-check. If your website says "Smith Home Loans", your Google Business Profile says "Smith Home Loans Pty Ltd", your aggregator listing says "J. Smith Mortgage Broking" and your MFAA entry has your old office address, the engine sees four weak entities instead of one strong one.
Pick one business name, one address format and one phone number. Then update, in this order: website footer and contact page, Google Business Profile, aggregator profile, MFAA or FBAA listing, LinkedIn company page, and any local directories you are already on. Do not create new listings for the sake of it. Fix the ones that exist.
Owned leads vs purchased leads
Brokers who buy leads know the problems: the same lead goes to several brokers, the person has often already spoken to a bank, and you are building nothing. Stop paying and the leads stop the same day.
Being recommended by an assistant works the other way. The enquiry is exclusive. The person arrives already trusting you, because the assistant vouched for you. And the asset compounds. A page that earns citations this year keeps earning them.
The honest trade-off is time. Purchased leads arrive tomorrow. Owned visibility takes months to build. One of our finance clients went from no organic visibility to being recommended as a top option across five AI platforms within seven months. That is a realistic horizon for a broker who starts now and does the work properly. Most brokers run both for a period, then taper the purchased leads as owned enquiries grow.
How to measure it each month
There is no rank tracker for ChatGPT the way there is for Google. The method that works is to ask the engines the questions your clients ask, and record what comes back.
Write down ten questions in your clients' words. For example: "best mortgage broker in [your suburb]", "first home buyer mortgage broker [your city]", "mortgage broker for self-employed near [suburb]", "should I use a broker or go to the bank for a refinance". On the same day each month, ask each question in ChatGPT, Gemini, Perplexity and a normal Google search with AI Overviews showing. For each answer, note three things.
| What to record | Why it matters |
|---|---|
| Were you named? | The outcome you are working toward. Track it per engine, per question. |
| Who was named instead? | These are the competitors to study. Look at what sources the engine cited for them. |
| What sources were cited? | Shows which directories, review sites and pages the engine trusts for your market. Get listed and accurate on those first. |
A spreadsheet with ten rows and four engine columns is enough. After three months you will see which fixes moved which engine. If you would rather have this done for you, our free AI visibility check runs exactly this process and tells you where you stand today.
Your 30-day checklist
Here is the work above as a plan. One person can do it alongside a normal workload. If you want it done faster and across more suburbs, that is what our AEO and GEO service is for. If you have colleagues in law, the law firm version of this guide follows the same structure.
| Week | Task | Done when |
|---|---|---|
| 1 | Ask the engines your ten client questions and record the results | Baseline spreadsheet exists |
| 1 | Check robots.txt and your CDN or firewall for AI bot blocks | GPTBot, OAI-SearchBot, PerplexityBot and Google-Extended are allowed |
| 1 | Add credit licence or credit representative number, licensee, aggregator and industry body to the site footer and about page | All four appear as plain text on every page |
| 2 | Settle on one business name, address and phone format | Website, Google Business Profile, aggregator profile and MFAA or FBAA listing all match |
| 2 | Build or fix a profile page for each broker | Photo, credentials, loan focus, suburbs and Person schema on each |
| 3 | Write your first loan type and suburb page | Direct answer in the first 120 words, FAQ, FinancialService and FAQPage schema, named author |
| 3 | Ask five recent clients for a Google review that mentions the loan type and suburb | Reviews posted and replied to |
| 4 | Write your second loan type and suburb page | Same standard as the first |
| 4 | Re-run the ten questions across the four engines | Month-two results recorded against the baseline |
None of this is clever. It is the discipline of being verifiable, local, specific and consistent, then checking whether it worked. Brokers who do it become the names the assistants give out. Brokers who do not keep paying for leads.
