When a buyer asks ChatGPT for the best real estate agent in their city, it names a median of six businesses. And 98% of the sources it cites to justify those picks are just two things: the agent's own website, and their Google Business Profile.
Zillow, Realtor.com, Redfin and Trulia — the portals that have defined online real estate discovery for twenty years — were cited zero times across 1,567 citations.
This report is a census of what AI search actually returns for real estate buying queries. Between April and May 2026 we captured 334 distinct ChatGPT answers across 253 buying queries in 197 property markets in the United States, the United Kingdom and the UAE, then analysed every business named and every source cited.
Omni Eclipse analysed 334 ChatGPT answers across 197 property markets. Four findings: (1) ChatGPT names a median of just 6 agents per city, and that number barely moves with market size. (2) 98% of the sources it cites are the agent's own website and Google Business Profile — Zillow, Realtor.com, Redfin and Trulia were cited zero times in 1,567 citations. (3) In US markets, 74% of those slots go to a named person or team, not a brokerage brand. (4) It justifies picks with countable proof — review counts, years in business, sales volume — not marketing copy. With 75% of buyers interviewing only one agent (NAR), the shortlist is close to the whole game.
Between April and May 2026 we ran natural buying queries through ChatGPT for each market — "best real estate agents in [city]" and "best luxury real estate agents in [city]" — and captured the full response text, including every link. The full corpus is 334 distinct answers covering 253 queries across 197 markets in the US, UK and UAE. Citation figures are calculated across all 334 answers (1,567 source links). Shortlist-size and agent-versus-firm figures are calculated on a 159-answer subset spanning 70 markets, where every one of the 976 businesses named was individually classified by hand. Brand-reach figures use the 254 answers that returned a structured list across 120 markets. Each figure below states its basis. This is a study of what AI search returns, not a ranking of agents.
How many real estate agents does ChatGPT actually recommend?
Six, typically. Rarely more than eight, and a third of answers give you five spots or fewer.
Across 159 answers where we classified every business named, ChatGPT returned a median of 6 businesses and a mean of 6.1, ranging from 3 to 10. Eight was the single most common answer shape, occurring 31.4% of the time. But 34.6% of answers listed five businesses or fewer.
This is the most important structural fact in AI search for real estate, and it does not behave like Google. A Google results page gives you ten organic listings, then a second page, then a third — a long tail that keeps going. Our data set includes markets where we captured firms ranked as deep as Google page 14. In AI search there is no page two. There is a shortlist of six, and everyone else does not exist in that answer.
| Businesses named | Answers | Share |
|---|---|---|
| 8 | 50 | 31.4% |
| 6 | 37 | 23.3% |
| 4 | 23 | 14.5% |
| 5 | 22 | 13.8% |
| 7 | 16 | 10.1% |
| 3 | 10 | 6.3% |
| 10 | 1 | 0.6% |
Adding "luxury" to the query does not buy you more room. Answers to "best luxury real estate agents in [city]" averaged 6.0 businesses against 6.3 for the plain query — slightly fewer, not more.
The scarcity compounds when you consider market size. The National Association of Realtors counts well over a million members in the US alone. A large metro has thousands of active agents. The shortlist is six.
And in real estate, making the shortlist is close to the whole transaction. NAR's Home Buyers and Sellers Generational Trends Report found that 75% of buyers interviewed only one real estate agent before choosing one (Exhibit 4-6), and 81% of sellers contacted only one agent before selecting who would sell their home (Exhibit 7-2). This is not a category where consumers assemble a comparison spreadsheet. Most people talk to one agent and hire them.
Stack those two facts together. Six names get surfaced. Three quarters of buyers interview exactly one. Being on the list is not a branding win — it is, in most cases, the sale.
Consumer adoption is already there. A Realtor.com survey of 1,000 US adults who were buying, selling, or had done so in the past two years found 82% use AI for housing market information, with ChatGPT the most-used platform at 67% (Realtor.com, Oct 2025).
Where does ChatGPT get its real estate recommendations from?
Almost entirely from two places, and neither is a property portal.
We extracted all 1,567 source links across the 334 answers, spanning 578 distinct domains. Google Business Profile and Maps listings accounted for 57.6% of citations. The business's own website accounted for 40.5%. Together those two surfaces are 98.1% of everything ChatGPT cited.
Local media and editorial coverage — the "best agents in [city]" listicles that agencies spend real money chasing — accounted for 1.4%. Real estate portals and directories accounted for 0.5%.
| Source type | Citations | Share |
|---|---|---|
| Google Business Profile / Maps | 903 | 57.6% |
| Own website (first-party) | 634 | 40.5% |
| Local media / editorial | 22 | 1.4% |
| Real estate portal / directory | 8 | 0.5% |
This aligns with what Yext found analysing 6.8 million AI citations: 86% come from brand-managed sources, split between first-party websites (44%) and business listings (42%) (Yext, Oct 2025). Our real estate figure is more extreme — 98.1% — but the shape is identical. AI search overwhelmingly cites surfaces the business itself controls.
That is the good news. Both of the surfaces that decide AI visibility in real estate are surfaces you own outright.
Does Zillow or Realtor.com influence AI recommendations?
On this evidence, no.
Across 1,567 citations we recorded zero links to Zillow, zero to Realtor.com, zero to Redfin, zero to Trulia and zero to Homes.com. Not a small number. Zero.
The portals do surface, but only as evidence about an agent inside the prose, never as a source. Zillow appeared four times as a rating reference ("a 5.0 rating on Zillow"). Redfin appeared eleven times, almost always as an employer ("Principal agent at Redfin"). In both cases the portal is being described, not cited.
Two decades of real estate marketing strategy assumed that portal presence equals discoverability. In AI search that link is broken. A polished Zillow profile with hundreds of reviews did not, in our data, generate a single citation. The surfaces that did generate citations were the agent's own website and their Google Business Profile. If your online presence is concentrated in the portals, you are invested in the one channel AI search did not read.
What evidence does ChatGPT cite when recommending an agent?
Proof of track record, not marketing copy.
We searched every answer for the kinds of justification ChatGPT attached to its picks. Review counts were the most common, appearing in 52.7% of answers. Years in business appeared in 47.6%. Sales volume in dollars appeared in 45.2%. Awards and rankings appeared in 42.2%. Star ratings appeared in 40.7%, and transaction counts in 27.2%.
The pattern is consistent: ChatGPT justifies real estate recommendations with countable, verifiable facts. "$230.9 million in sales volume." "4.9 rating from 649 reviews." "Over 3,100 sales in Charlotte." These are numbers that live on Google Business Profiles and on agents' own about pages — which is precisely where the citations came from.
Descriptive marketing language — "trusted", "premier", "your local expert" — is not what gets quoted back. Specific numbers are.
Does a national brokerage brand get you recommended?
Much less than you would expect.
Of the 1,550 businesses named across 254 structured answers in 120 markets, only 14.7% carried a national brokerage brand in their name. The rest were independents, local teams, and individual agents.
The national brands do have the widest footprint — Keller Williams appeared in 28 markets, Sotheby's in 26, Coldwell Banker in 21, Compass in 18 — but that is reach across markets, not dominance within them. In any single city answer, brand names are a minority of the shortlist.
More striking: of the 1,446 distinct businesses named, 99.4% appeared in exactly one market. AI search for real estate is almost perfectly hyperlocal. A recognised national brand does not travel into the answer; being the best-documented operator in one specific city does.
| Brand | Markets named in |
|---|---|
| Keller Williams | 28 |
| Sotheby International Realty | 26 |
| Coldwell Banker | 21 |
| Compass | 18 |
| eXp Realty | 15 |
| RE/MAX | 13 |
Does ChatGPT recommend agents or brokerages?
People, overwhelmingly. This is the finding most likely to be aimed at the wrong target by brokerage marketing budgets.
We hand-classified every one of the 976 businesses named across 159 answers. Just under half — 49.8% — were a single named individual agent. Another 24.1% were a named agent team. Only 22.8% were a brokerage or agency brand. Put together, 73.9% of every slot went to a person or a named team rather than a company.
The pattern holds at the top of the list, too: the first business named in an answer was an individual agent 50.3% of the time. And 40.9% of answers named no firm at all — not one. Only 6.3% consisted purely of brokerages.
That matters because "best real estate agents in [city]" is a question about people, and ChatGPT answers it literally. A brokerage investing entirely in firm-level brand while its individual agents have thin personal footprints is competing for less than a quarter of the shortlist.
There is one important exception. This is a US pattern. In the UK and UAE markets we sampled, the identical prompt template returned agency brands almost exclusively — 97.9% firms in the UK, 100% in the UAE. Those are small samples (48 and 8 slots respectively) and should be treated as directional, but the divergence is stark enough to matter for anyone marketing outside the US.
The query wording also shifts the answer. Plain "best real estate agents" queries were the most individual-dominated at 63.8% individuals, while "best luxury real estate agents" dropped to 38.2% individuals and leaned toward teams and firms.
The practical read for US agents: you need your own documented presence — your own site, your own Google Business Profile, your own review history and your own verifiable numbers — not just a profile page on the brokerage site.
“Across 334 AI answers, 98.1% of every source cited was either the business's own website or its Google Business Profile. The two surfaces that decide AI visibility in real estate are the two surfaces you already control.
What should real estate agents do about this?
The findings point to an unusually clear action list, because the surfaces that matter are ones you own.
Treat your Google Business Profile as primary infrastructure. It was the single largest citation source at 57.6%. Complete every field, keep hours and service areas accurate, and build review volume deliberately — review counts were the most frequently cited justification in the entire study.
Publish your numbers on your own website. Sales volume, transaction counts, years in the market, awards, review totals. ChatGPT quoted these facts constantly, and 40.5% of its citations pointed at business-owned sites. If your about page has adjectives where competitors have figures, you have given the model nothing to cite.
Build presence at the individual agent level, not just the brokerage. In US markets, 73.9% of shortlist slots went to a person or a named team, and only 22.8% to a brokerage brand.
Go deep in one market rather than broad across many. 99.4% of named businesses appeared in exactly one market. Local dominance is what registers.
Do not assume portal presence covers you. It did not generate a single citation in this study.
This is the same underlying discipline we cover in how to rank in ChatGPT and how to get your business recommended by ChatGPT, applied to a vertical where the citation pattern is unusually concentrated. For the cross-industry picture, see our 2026 AI Search Visibility Report and State of AI Search 2026. Agents comparing channels may also want our AEO vs SEO breakdown and our guide to AI search for real estate agents.
The timing matters. McKinsey found 44% of consumers now prefer AI search for buying decisions, against 31% for traditional search. Bain & Company found 80% of consumers rely on AI results for at least 40% of searches. Gartner projects traditional search volume falling 25% by the end of 2026.
For a business where a single transaction can be worth tens of thousands in commission, six shortlist slots per city is a high-stakes number.
Want to know if you're in the six?
Book a free AI Visibility Audit. We'll check whether you appear in ChatGPT, Perplexity, Gemini and Google AI Overviews for your market, and show you exactly which signals are missing.
Book Your AI Visibility AuditFrequently Asked Questions
How did you collect this data?
Between April and May 2026 we ran natural buying queries through ChatGPT for each market, using the phrasing a real buyer would use: "best real estate agents in [city]" and "best luxury real estate agents in [city]". We captured the complete response text including all links, then parsed each answer for the businesses named and the domains cited. The full corpus is 334 distinct answers across 253 queries and 197 markets in the US, UK and UAE, and citation figures use all of it. Shortlist-size and agent-versus-firm figures use a 159-answer subset covering 70 markets in which all 976 named businesses were classified by hand; brand-reach figures use the 254 answers that returned a structured list across 120 markets.
Why does ChatGPT name only six agents?
Answer length is the constraint. ChatGPT returns a conversational shortlist, not a paginated index, and in our data that shortlist ran from 3 to 10 businesses with a median of 6 — regardless of how large the market was. A major metro does not get a meaningfully longer list than a small one. This is a structural difference from Google, where a deep long tail of results remains accessible to searchers who keep scrolling.
Does this mean Zillow and Realtor.com no longer matter?
Not at all — they remain major sources of direct consumer traffic and leads. What our data shows is narrower and specific: they were not cited as sources when ChatGPT justified a real estate recommendation. Portal presence and AI search visibility are separate channels driven by different signals. Strength in one does not transfer to the other.
Which AI platforms did you test?
This study covers ChatGPT specifically, which holds the largest share of AI search queries and consumer buying behaviour. Perplexity, Gemini and Google AI Overviews use different retrieval methods and cite different source mixes, so these figures should not be assumed to generalise across platforms.
Is this a ranking of the best real estate agents?
No. This is a study of what AI search returns, not an assessment of any agent's quality. Businesses named in ChatGPT's answers appear here as data points about the model's behaviour. Appearing in an AI answer reflects documented, machine-readable presence — it is not an endorsement, and absence is not a judgement of quality.
How stable are these results?
AI search visibility is considerably less stable than Google rankings, with BrightEdge reporting substantial monthly citation churn in ChatGPT. Our data is a snapshot from April to May 2026. The structural findings — shortlist scarcity, and citation concentration in owned surfaces — are the durable part; the specific businesses named in any given market should be expected to change.
Ashur Homa
Built and scaled a digital brand to $100M+ in sales with zero ad spend. Has helped businesses generate millions through AI go-to-market strategy. Leads growth at Omni Eclipse.
Connect on LinkedIn
