Real estate in AI search: showing up in the answer.
Buyers are asking ChatGPT which agent to call before they ever open Zillow. Here is how answer engines pick a real estate team, and the playbook for becoming the one they name in your market.
By Anna Curry
13 min read
July 2026
A family three states away has decided to move to your market. They do not know a single person there. Ten years ago they would have opened Zillow and clicked a photo. Five years ago they would have Googled “best realtor in [city]” and skimmed a page of blue links.
Today a growing share of them open ChatGPT and ask a question. And ChatGPT gives them one answer, sometimes two. Not ten. Not a page of options with your ad above it. One name, presented as a recommendation from something that sounds like it knows.
That is the whole shift. Search used to be a list you could buy your way onto. The answer is a recommendation you have to earn. This guide is about how real estate teams earn it.
Why the biggest brokerage usually loses.
Here is what surprises most agents: answer engines are not impressed by size. A three-agent boutique routinely gets named ahead of a franchise with a national ad budget, and the reason is structural.
Franchise marketing is designed to be broad. The same brand language, applied across hundreds of markets, deliberately non-specific so it works everywhere. That is exactly the kind of copy an answer engine has nothing to grab. When a model is asked who knows the Wrightsville Beach short-term rental market, “trusted since 1972, nationwide” does not answer the question.
Specificity does. A team that has published a real school-district guide, holds a local award with a page attached, and has verified transaction data in its market gives the engine something quotable. AI recommends whoever the evidence describes most clearly, and small teams can be far more specific than large ones.
Answer engines do not reward the biggest brand. They reward the clearest entity.
The prompts running in your market right now.
Before any strategy, you need the questions. Every market has a working set of prompts buyers and sellers are typing, and almost no team has ever read theirs. They look like this:
Notice what they have in common. None of them are keywords. They are the questions a person would ask a knowledgeable friend, and they carry qualifiers: out of state, first time, short-term rental, a specific neighborhood. Each qualifier is a separate answer, and most of them are currently unowned in your market.
That is the opportunity. You do not have to win “best realtor in [city]” on day one. You have to find the high-intent question nobody has claimed and become the obvious answer to it.
Four moves, in order.
The sequence matters more than the individual tactics. Teams that skip to step four wonder why nothing sticks.
Read your market’s prompts.
Find out what buyers actually ask ChatGPT, Perplexity, Gemini, Google AI Overviews, and AI Mode about your market, and who gets named today. Most teams are surprised by both lists: the questions they never considered, and the competitor quietly winning them. Everything downstream is prioritized off this.
Fix the foundation.
One domain, not three. Agent bio pages built as entities rather than brochures, each with credentials, market focus, and consistent naming. A CRM that captures every inquiry. Reviews with real responses. Structured data that tells engines what your pages are.
This is the unglamorous year. Everything later depends on it.
Make the evidence agree.
Answer engines do not take your word for anything. They triangulate across sources they already trust, and in real estate that means marketplaces, transaction-data profiles, verified rankings, local press, and awards with a published page behind them.
The job is to make the same facts appear, phrased consistently, everywhere the engines read. Contradictory or missing evidence is why capable teams stay invisible.
Take a high-intent question, then expand.
Pick the question with real buying intent and the weakest incumbent. Relocation is often the best first target: those buyers have no local network, the highest need for guidance, and the longest research window.
Then build what deserves the citation. A cornerstone page, a genuinely useful downloadable guide, and an ongoing publication that keeps the topic alive. One answer owned properly beats six half-claimed.
Content that gets quoted.
Most real estate content is written to look active rather than to be cited. Listing announcements, market-stat graphics, and holiday posts give an engine nothing to quote, because they answer no question a buyer asked.
What does get quoted is specific, structured, and durable. Neighborhood guides that name streets, commute times, and flood zones. Relocation guides that address the real anxieties of moving sight unseen. Guides to the practical questions every buyer asks, built with the local professionals who can actually answer them: a lender on financing, an inspector on older housing stock, a contractor on what renovation costs here. Buying and selling walkthroughs that explain your market’s particular quirks. Recurring formats that give the topic a heartbeat rather than a one-off post.
The test is simple: could a model quote one sentence from this page as the answer to a question a real buyer typed? If not, it is marketing, not evidence.
If a model cannot quote a sentence from the page, it is marketing, not evidence.
What to expect, honestly.
As an agent, here is what you need to know about pacing. The substrate work takes months and shows almost nothing while it is happening. Third-party proof compounds slowly, because you are waiting on awards cycles, ranking submissions, and data refreshes you do not control.
Then it moves quickly. Once the evidence is in place and you publish something genuinely useful on a high-intent topic, AI answers can shift in weeks rather than quarters. In the engagement below, visibility on the relocation prompt roughly tripled in six weeks once the guide, the cornerstone page, and the supporting citations landed together.
This is not deterministic the way a Google ranking is. The same question can return different sources from one request to the next. The work improves your eligibility to be cited and the consistency with which you appear.
A three-agent team that became the relocation answer.
Gillespie Group Real Estate in Wilmington, NC ran exactly this playbook over three and a half years. Today they hold the number one AI visibility position for relocation to their market, ahead of every legacy brokerage, and their own website is the third most cited source for Wilmington real estate answers.
Read the case studyQuestions agents ask.
How do AI engines decide which real estate team to recommend?
They synthesize evidence from sources they already trust: real estate marketplaces, transaction-data and ranking sites, local publications and awards, review platforms, and the team’s own website. When the same facts about a team appear consistently across those sources, the engines treat the team as the credible answer. Answer engine optimization is the work of building and aligning that evidence.
Is this different from real estate SEO?
It overlaps but the target changed. Traditional SEO optimizes for a ranked list of links, which rewards keyword coverage and backlink volume. Answer engine optimization targets a synthesized recommendation, which rewards entity clarity, structured data, third-party corroboration, and content specific enough to quote. Good SEO helps. It is no longer sufficient on its own.
Do I have to be a big brokerage for this to work?
The opposite is often true. A focused team with real production, real reviews, and a disciplined content engine can out-cite a franchise, because answer engines reward specific, consistent, well-evidenced entities over generic brand size.
Does this only work in smaller markets?
No. It works anywhere buyers ask AI for a recommendation, which is every market. Larger metros have more competitors, but also more prompts, more niches, and more high-intent questions no team currently owns. The playbook does not change: read the prompts, fix the substrate, stack the proof, own one answer, then expand.
How long before we show up?
The foundation takes months and the third-party proof compounds slowly. Movement can be fast once the evidence lands: in a documented engagement, visibility on a target prompt roughly tripled in six weeks after the cornerstone content and citations were in place. The full arc from invisible to market leader across many prompts took three years.
Where should a team start?
With the prompts. You cannot prioritize anything until you know what your market is asking and who is being named today. That is what the AI Visibility Audit produces, and it is the input every other decision depends on.
Anna Curry is the founder of Top of Search, a marketing consultancy in Wilmington, NC that helps owner-operated local businesses get cited by AI search engines. She has worked in marketing since 2008 and writes Cited, a weekly newsletter on AI visibility for local business owners.
More about AnnaYour market has its own set of prompts.
Find out what AI says when buyers ask about your market.
The AI Visibility Audit is where every engagement starts. We run the real prompts for your market across five answer engines, score where you stand against the teams you compete with, and hand you the evidence, whether or not we ever work together.