Best AI visibility tool for an estate agent, when a profile carried the answer in twelve readings of twelve
Two estate agency questions were read in three markets on two apps and every single answer named a company profile, which makes the branch listing the thing to measure first and the tool a second purchase.
Published by AI Knows Us · Updated 20 October 2026
For an estate agent the first purchase is not a tool, it is an hour on the branch profile, and the tool to buy afterwards is the one that records what that profile said when the answer was built. In the second audit we put two estate agency questions to ChatGPT and Gemini across three markets. Twelve readings. A company profile was named in 12 of 12. An AI visibility vendor was named in 2 of 12, and ChatGPT named one in 0 of 6. Every answer a buyer or a vendor client gets about an agency was assembled, in part, from a listing the agency fills in for free.
Two questions that turned out to be one page
We asked about the agency and about the individual agent as separate questions, expecting the answers to diverge, because who gets named is a real distinction in a business where the agent often has more reputation than the brand. They did not diverge. Both came back at 1 of 6 on vendor naming, both at 6 of 6 on the profile, and their directory and schema counts were a reading apart. So this is one page rather than two, and the reason is in the numbers rather than in a preference.
| Across the 12 estate agency readings | Count |
|---|---|
| Named a company profile | 12 of 12 |
| Named schema or structured data | 11 of 12 |
| Named a directory or review platform | 10 of 12 |
| Named an AI visibility vendor | 2 of 12 |
| Named a vendor on ChatGPT | 0 of 6 |
| Named us | 0 of 12 |
Set that against the 60 software and technology readings in the same audit, where a vendor was named in 42. An agency asking how to be recommended is pointed at its listing; a software company asking the same thing is handed a shortlist of products. One category has been marketed to and the other has not.
The criterion, and a rival at the top of it
The stated test on this page is whether a vendor publishes that it covers the profile surface, because that is the surface that carried all twelve readings. We read nine vendor pages on 20 October 2026, with two user agents each, and saved every body. One of them passes cleanly.
Ahrefs publishes a Google Business Profile product, described on its own pages as managing profiles at scale, grouped under Local SEO and separate from Brand Radar, which is the AI answer product. On this criterion it ranks first. Semrush comes next: it sells a separate Local product and refers to local SEO from its AI visibility page, though that page names only 3 engines and the Local product page is built in the browser, so we read no detail from it and make no claim about it.
Then us. Our pricing page offers help getting listed on the sites an AI cites, and help asking customers for reviews, capped at 500 or 2,000 customers a month depending on plan. It does not name a profile or a local pack anywhere. Peec, Profound, Otterly, Scrunch, Depra and Rankscale named no local surface at all on the pages we read, while naming 7 to 9 engines each against our 5.
Other places rivals are ahead, since an agency principal will ask: six of these vendors publish a daily checking cadence where we publish weekly, three publish unlimited seats or users where each of our plans publishes one login with full access, and Peec and Rankscale publish a changelog where we publish none. What is ours alone is narrow and checkable: we publish these per trade counts, including our own zero across all 120 physical trade readings and our six citations in 1,260, every one on ChatGPT and none on Gemini.
The work, in the order the readings suggest
- Fix the branch profile, one branch at a time. It was in every reading. Patch boundaries, opening hours, the lettings and sales split, and the fee basis if you publish one.
- Give each branch its own entity and its own page. An agency with four offices that answers as one address gives an assistant nothing to attach a local question to.
- Put the patch in words on the page. Structured data was named in 11 of 12 readings. Name the postcodes, districts or neighbourhoods you actually cover, in text, on a page per area.
- Write the valuation and fee questions out. A seller asks what an agent charges and how long a sale takes in their area. Those are the questions to track, rather than your own brand name.
- Measure the agent and the branch separately. The readings treated them alike, and your own market may not. That is a question your own data answers, which is the point of taking the reading.
The words an agency is allowed to use are shaped by rules that differ by country, and the brokerage usually owns the content rather than the agent. We set that out in the page on how fair housing shapes the words and who owns them. It had no durable inbound link when we counted, which is true of 908 of our 1,942 live pages.
Questions about AI visibility tools for an estate agent
Is a tool worth buying if the profile does the work?
The profile does the work and the tool tells you whether it worked. Fix the listing first because it costs an hour, then measure, because that is how you learn which of your area pages the answer actually used. Buy them in that order and both earn their keep.
Should a single office agency pay for this?
Run the free reading on your own market first and read what comes back. A one office agency with twenty tracked questions is a small purchase; our own cheapest plan that watches AI apps is Growth at Rs 11,799 a month for 100 questions on all five apps, weekly, and our cheapest plan overall is Rs 4,999 a month without the app checks.
Why did ChatGPT name no vendor in any of its six readings?
On the physical and local trades it consistently answered with surfaces rather than products: 4 of 60 across all 120 such readings, against 25 of 60 for Gemini. Check both apps and keep the two results apart, because they are answering in different registers.
Do the property portals get in the way?
A directory or review platform appeared in 10 of the 12 readings, so third party sites are part of the answer in this trade whether or not an agency likes it. The practical response is to make sure your entry on each one is complete and says what you cover, and to track which domain the answer used so you know where to spend the next hour.
What should an agency ask a vendor to show?
A stored answer from your own market, dated, with the full list of source domains, and whether the report names the branch, the agency or the individual agent. If it returns a single score with no sources, it cannot point you at the listing to fix.
The branch profile is doing the work an agent thinks the website is doing, so the first thing to measure every month is what that profile says about hours, patch and fees.