What AI says about real estate agents
ChatGPT, Gemini and Google's AI Overviews answer questions about real estate agents every day, and the people asking rarely double-check. Here is what they get wrong about an estate agency, why, and how to find out what yours says.
What is at stake for real estate agents
Sellers shortlist two or three agents before making a single call, and that shortlist is increasingly assembled by an engine. Being absent from it is invisible — there is no missed call to notice.
The awkward part is that none of this shows up as a problem you can see. There is no bounced call and no complaint — the customer asked, got an answer, and went somewhere else. The first sign is usually a quiet month you cannot account for.
What customers actually ask
These are the questions being put to engines about real estate agents, and they are worth reading closely: none of them is your business name on its own. People ask about a need, and the engine decides who fits.
- “Which estate agents cover [neighbourhood]?”
- “What commission does [agency] charge?”
- “Does [agency] do lettings as well as sales?”
- “How quickly does [agency] sell properties in [area]?”
- “Best estate agent in [town] for selling a flat?”
Ask each of those about your own agency, from a private window, and read what comes back. That is the audit — there is not a more sophisticated version of it.
The errors that specifically bite real estate agents
Every trade has its own failure modes, and generic advice misses them. These are the ones that recur for real estate agents:
- Coverage area understated, so neighbouring postcodes never see you
- Fee structure wrong or missing, which is the first thing sellers filter on
- Lettings capability not mentioned when you offer it
- Branch details wrong after an office move or merger
The listings that matter here
Google Business Profile is the foundation for everyone, but the listings that carry weight after it differ sharply by trade. For real estate agents these are the ones worth auditing:
- Google Business Profile
- Rightmove and Zoopla, or Zillow and Realtor.com, which are read as authoritative on coverage
- Your franchise or network’s branch finder
- GetAgent or similar comparison sites
- Apple Maps
Reviews for real estate agents accumulate mainly on Google, Trustpilot, Zillow or Rightmove, Facebook — and the text of those reviews is being read, not just the star average. A model asked for the best option for a specific need can only name you if somebody wrote about that specific need.
The one thing most real estate agents get wrong
Name the neighbourhoods you cover individually rather than describing a radius. Sellers ask about the area they live in by its name, and a page saying "covering greater [city]" matches none of those questions, while a list of twelve named neighbourhoods matches twelve of them.
How to check yours
- Open a private window. In ChatGPT, turn memory off — otherwise it may be answering from something you told it months ago.
- Ask the five questions above, using your full street address so you are not reading an answer about a business with a similar name.
- For anything wrong, ask “what sources did you use?” Fix the page it names. If it cannot name one, the detail was invented and the fix is making the correct fact easier to find.
- Paste the answers into a document with today’s date. A single audit tells you about today; what you need to know is whether things improve after you change something.
- Check again in a month. Corrections propagate on their own schedule.
The longer version of this process, with the exact prompts, is in how to check what AI says about your business.
If you would rather not do it by hand
Once a month by hand is fine for one location that rarely changes. It stops being practical when you have several, or when you need to know the week something breaks rather than the month.
AI Business Monitor runs these questions weekly for your agency, stores every answer so you can see what changed, and emails you when something moves — with the source behind each finding. It cannot change what an engine says, and neither can anyone else; what it does is tell you what is being said.
Related reading
- AI says my business is closed — the most expensive error there is, and the one to fix first.
- Wrong hours, phone or address — tracing each error back to its source.
- Getting recommended by AI — what actually influences whether you are named.
Common questions
How do I find out what ChatGPT says about my agency?
Ask it directly, in a private window with chat memory turned off so it is not answering from things you told it before: "What do you know about [name] at [full street address]? Include hours, services and contact details." Then ask "what sources did you use?" — the page it names is usually the thing that needs fixing.
What does AI most often get wrong about real estate agents?
Coverage area understated, so neighbouring postcodes never see you. Fee structure wrong or missing, which is the first thing sellers filter on. Both come from a source that was correct once and was never updated, which is why finding the source matters more than correcting the engine.
Can I make an AI engine correct itself?
Not directly — there is no listing to edit and no correction process, and nothing typed into a chat window teaches a model anything about your business. What works is fixing the underlying sources it reads, then waiting. Google Business Profile changes often show up within days; ChatGPT can take weeks.
Which listings matter most for real estate agents?
Google Business Profile first, always. After that the ones specific to this trade: Rightmove and Zoopla, or Zillow and Realtor.com, which are read as authoritative on coverage, Your franchise or network’s branch finder, GetAgent or similar comparison sites. Trade-specific directories carry more weight for these questions than general ones, and they are the listings owners most often forget exist.
Check it every week automatically
Doing this by hand once tells you about today. Things change: hours get edited, reviews land, a competitor opens. We check weekly and email you when something moves, with the source behind every finding.