How to Rank in ChatGPT: What 2,000 Real Estate Answers Show

SparkToro has already shown that ChatGPT rarely repeats the same list of brands, so a share of answers is a better measure than a position. This guide covers what that point leaves out: how ChatGPT handles city-level questions, how often engines disagree with each other, why your own test can mislead you, and which common tactics have any evidence behind them.

Can you actually rank in ChatGPT?

Not in the sense of a fixed position. ChatGPT writes each answer fresh, and SparkToro measured under a 1 in 100 chance of the same brand list appearing twice. Engines also disagree with each other: on 432 Houston real estate questions, ChatGPT and Perplexity agreed on only 5.1% of the firms either one named.

SparkToro’s study with Gumshoe.ai, published in January 2026, collected 2,961 responses from ChatGPT, Claude and Google AI and concluded that a visibility percentage across many runs is “a reasonable metric” while a ranking position is not. Gumshoe sells AI tracking, which the post discloses.

Our Houston data adds the cross-engine view. On the 432 questions that ChatGPT, Perplexity and both versions of Gemini all answered, ChatGPT named 803 distinct firms, and only 133 firms, 5.1% of the combined ChatGPT and Perplexity list, were named by both. Two assistants given identical questions in the same week mostly recommended different people, so in practice how to rank in ChatGPT means raising your share, question by question and place by place.

Each square is one firm, drawn to scale. Source: CiteVantage Houston study, September 2026.

Share of answers is the percentage of runs, across a fixed set of questions, in which an AI assistant names your business. It is the measure of AI search visibility that holds up when no stable position exists.

How do you rank in ChatGPT search?

Let OpenAI’s search crawler in, then be worth citing. OpenAI says sites that block OAI-SearchBot will not be shown in ChatGPT search answers, and it recommends allowing it. GPTBot is a separate setting that controls training only. After you change robots.txt, OpenAI says search takes about 24 hours to adjust.

OpenAI describes the search side in its launch post for ChatGPT search, dated 31 October 2024: “ChatGPT search leverages third-party search providers, as well as content provided directly by our partners, to provide the information users are looking for.” It does not say which providers, or how it picks among the pages they return. Ahrefs found, across 118,931 fan-out queries, that ChatGPT ran an average of 1.78 searches per prompt.

The crawlers are where OpenAI is specific. Its crawler documentation says “each setting is independent of the others”:

Robots.txt is only half of access, since a host can refuse a crawler your file allows. When we tested our own site on 14 September 2026, our host returned HTTP 429 to 8 of 8 cache-miss requests carrying the versioned GPTBot user agent, while OAI-SearchBot passed on 8 of 8. Search access was intact, because only the training crawler was refused, and the full trace is in our GPTBot case study. Test each OpenAI user agent separately, using the steps in our guide on whether AI crawlers can reach your site.

How does ChatGPT choose who to recommend in your city?

It names very few firms, and it leans on property portals to pick them. In our study of 2,000 AI answers to 500 Houston real estate questions, 94.0% of the Houston Association of Realtors’ 50,000-plus members were never named. ChatGPT sent 48.7% of its citations to portals and named the fewest firms of the four surfaces.

How four AI surfaces answered the same Houston real estate questions, September 2026

Source: CiteVantage, 500-Query Houston Real Estate AI Visibility Study, fieldwork 18 to 22 September 2026, one run. Row-level data is on the study page, and journalists can find a summary on our press page.

Perplexity named more than twice as many firms as ChatGPT on the same questions. Of ChatGPT’s 599 citations, 48.7% pointed at portals and another 11.0% at agent referral networks, so most of the sources behind its picks were sites the agents do not own. ChatGPT also named no firm at all in 43 of its 453 geo-clean answers, or 9.5%.

One square is one ChatGPT citation. Source: CiteVantage Houston study, September 2026.

The pattern holds nationally. Local Falcon’s Realtor AI Visibility Index, a vendor measurement of 37,500 “best realtor near me” style queries across the 100 largest US cities in August 2026, found zillow.com accounted for 39.8% of ChatGPT’s sources and realtor.com for 18.0%.

Why does ChatGPT show you a different answer from your clients?

Because ChatGPT tailors its searches to the person asking. OpenAI’s help center says it collects general location from your IP address and, with memory on, may use saved memories when it rewrites a search. In our Houston study, questions without a state marker were answered about the wrong place 4.5% of the time, against 1.1% with one.

OpenAI’s article on searching the web with ChatGPT puts it plainly: “ChatGPT also collects general location information based on your IP address and may share that general location with third-party search providers to improve the accuracy of your results.” It adds: “If memory is enabled, ChatGPT may use relevant saved memories when rewriting a search query.” An agent checking from their own office, on their own account, can get a different answer from a buyer relocating from another state.

Place wording does measurable work. Houston queries carrying a marker such as “TX” went to the wrong place in 10 of 880 answers, against 50 of 1,120 without one. The errors were also sticky: of 37 ChatGPT wrong-place answers asked again two days later, 29 were still wrong, and 27 of those named the same wrong place.

One square is one answer. The two groups of questions are compared across the full set, not as matched pairs. Source: CiteVantage Houston study, September 2026.

To see what a buyer sees:

• Keep a fixed list of questions, worded the way a buyer in your area would type them.
• Run each one several times in a fresh chat with memory off, never from the account you use every day.
• Write the city and state into every question, for example “Katy, TX”.
• Record every firm named and every domain cited, then divide the runs that name you by the total.

Our post on how to check if ChatGPT recommends your brand has a scored version of this test.

Does ranking on Google, or being a top producer, get you named?

Neither guarantees it. Ahrefs found only 8.0% of ChatGPT’s in-text citations ranked in Google’s top 10 for the same question. Local Falcon found only 21.2% of the highest-volume agents with a website were cited by any AI platform. Seer Interactive, though, found Google page-one rankings correlated with being named in GPT-4o answers.

The Ahrefs study covered about 15,000 long-tail queries collected in July 2025. The 12% figure often quoted from it is the average across ChatGPT, Gemini and Copilot, so it overstates ChatGPT.

Local Falcon’s sales-volume check is the one closest to an agent’s worry. Of the top 10 individual agents per city by RealTrends Verified 2026 volume who publish a website, 106 of 499 were cited by any AI platform. On that evidence, closing more deals than anyone in your city does not reliably get you named.

On the other side, Seer Interactive’s study of 10,000 questions in finance and SaaS found brand mentions in GPT-4o answers correlated about 0.65 with Google page-one rankings, while backlinks looked “weak or even neutral”. That used the API model, not the ChatGPT app, and a correlation does not show cause. The two findings can both hold: Google rankings may help you get named even when they do not decide which page gets linked. Keep doing SEO, and do not expect it to carry you into the answer alone.

Site strength is a separate question, and our own check points the same way. In our 150-question study of AI search for real estate, we pulled Ahrefs Domain Rating and estimated organic traffic for a random 200 of the 1,197 cited domains outside the 100 most-cited. The median Domain Rating was 32 and 32% sat under 20. Some 46% drew fewer than 100 organic visits a month, and 27.5% had both a Domain Rating under 20 and fewer than 100 visits (Ahrefs, 29 October 2026). A weak SEO base did not keep those sites out of the answers. The 100 most-cited domains had a median Domain Rating of 52, so strength still helps. The questions were about hiring real estate and AI SEO providers, and many cited sites are new agency and tool sites, so the result may not carry over to local agent questions.

How to rank in ChatGPT: what is proven and what is guesswork?

Very little is proven. OpenAI documents its crawlers and says ads do not affect answers. Everything about how sources are picked comes from outside measurement, and most of that is correlation from companies that sell tracking tools. Of the common tactics, only crawler access is documented by OpenAI itself.

Claims about ranking in ChatGPT, sorted by evidence

The AirOps figure comes from more than 500 commercial SaaS queries, with 13.2% of mentions from the brand’s own site. The last row matters most to a busy agent, since those are the tactics most often sold as quick fixes.

How to rank in ChatGPT for your city: what raises your share?

To rank higher in ChatGPT for your area, get named on the pages it already reads for your city, and be exact about place. In Houston, a firm’s own website was cited in only 8.4% to 10.4% of the answers naming it, and 82.6% of the real agents, teams and brokerages named appeared in one of 40 areas.

Local Falcon reported another pattern among the agents who do get cited. Eighteen of the 25 most-cited agent websites carry a “best real estate agents in [city]” page, and sites with such a page took a third of every citation that went to an agent website. Local Falcon measured which sites get cited, and did not test whether the page caused it.

What we would work on, in order:

1. Allow OAI-SearchBot in robots.txt and confirm your host serves it, deciding on GPTBot separately.
2. Portal and referral-network profiles. Complete every one you hold, since that is where 48.7% of ChatGPT’s Houston citations went.
3. One page for each area you serve, with the city and state written into the copy.
4. Page titles that repeat the buyer’s question in their words. Ahrefs reported that cited pages had titles closer to the prompt than pages ChatGPT retrieved and skipped, using an approximate similarity measure.
5. Mentions on the sites that already appear in ChatGPT’s sources for your city, such as local news, best-agent roundups and community forums.
6. Your name, service area and description kept identical on every profile, so each mention points to one business.

The page-level craft, meaning answer blocks, schema and entity markup, is in what it takes to get cited by ChatGPT. For agents, our guide to AI search for real estate agents sets out the five signals and a 30-day plan. In our experience the slow part is authority, meaning other sites choosing to mention you, and no study has shown on-page work alone fixes visibility.

Can you pay to rank in ChatGPT?

Not inside the answer, according to OpenAI. Ads may appear for users on the Free and Go plans, while Plus, Pro, Business, Enterprise and Edu accounts do not show them. OpenAI says ads do not influence ChatGPT’s answers, which is its own statement, and we know of no independent audit of it.

OpenAI’s help article on ads in ChatGPT says the ads are being rolled out gradually in select regions, and states: “Ads do not influence ChatGPT’s answers.” Its crawler page lists OAI-AdsBot for checking ad landing pages, which is the only part of the ad system a website owner touches. The organic recommendation has no paid entry point, so being named there still comes down to access, pages worth citing and mentions elsewhere.

See your own share of answers

Any plan for how to rank in ChatGPT starts with the actual answers for your area. Our free AI visibility audit runs the questions your buyers ask and sends back, within 48 hours, the answers with screenshots, the firms named instead of you, and what to fix first. For the wider picture, see our real estate AI SEO services.

Other engines work differently: see our guides to getting cited by Perplexity and getting cited by Gemini.