The Realtor AI Visibility Index

Ask ChatGPT who the best real estate agent in your city is and you get a shortlist of names. Ask Google's AI Overviews the same question and, seven times out of ten, you get nothing at all. Ask all five major AI platforms and they will barely agree on a single name between them.

Real estate has never lacked for rankings. Portals publish them, brokerages publish them, trade titles publish them. What nobody has measured is which names the AI systems themselves return when a buyer asks, and whether those names belong to the agents actually selling the most homes.

Local Falcon ran 37,500 AI searches across the 100 largest US cities to find out which real estate agents artificial intelligence is willing to name, which ones it picks, and where those answers come from.

Key findings:

How Short the Shortlist is

Across ChatGPT, Google AI Overviews, Gemini and Grok, 3,757 distinct agents or teams were named in the 100 largest US cities. A typical city produced 42 of them.

No public source counts how many agents work inside the areas scanned, so the visible field is measured against a defined benchmark instead. RealTrends Verified ranks agents and teams by production in these markets. Of those who publish a website, AI cited 8.5%.

The visible field is not only small. Within each city, it is dominated by a handful of names. The table below measures concentration inside each of the 100 cities separately, then averages those results.

Share of a city's named agents | Typical number of agents | Average share of that city's agent citations

In the average US city, a single agent takes more than a quarter of everything AI says about agents in that market, and four agents take more than half. The pattern holds everywhere: the most-cited agent claims at least 7% of their city's citations in the flattest market and 51% in the most concentrated.

The Agents AI Recommends are not the Agents who Sell Houses

Four in Five Top Producers are Invisible

RealTrends Verified publishes annual production rankings for US agents and teams, measured in sales volume and transaction sides.

Cross-referencing its 2026 rankings against every agent website cited in this study gives a direct answer to whether AI recommends the people who actually sell homes.

It does not.

The reverse direction says the same thing. Nearly nine in ten of the agents AI recommends hold no production ranking in their own city, by volume or by transaction sides.

A meaningful minority of top producers have solved it, which shows the problem is not structural. Boston's highest-volume agent, Tracy Campion, is cited. So are Katrina Barrett in Phoenix and Kumara Wilcoxon in Austin, each the top individual seller in their market. Among teams, the Stephanie Younger Group in Los Angeles and the Mike McCann Team in Philadelphia both rank first in their cities and both get cited.

The Winners Publish a Best-Of Page and Name Themselves the Best

Looking at what is on the most-cited agent pages explains the mismatch. These are the pages behind the highest-cited agent domains in the study:

These pages are not neutral comparisons. The agent who publishes the page is almost always the one named best on it, and AI treats the page as the authority on who is best.

The pattern holds at scale. Agent sites carrying a ranking page of this kind make up a small minority of those cited, yet they absorb a third of every citation an agent website received. Among the 25 most-cited agent sites, 18 carry one.

Production volume is invisible to a language model. A published ranking is not. Agents who write the ranking get cited; agents who top the ranking do not.

Compass Leads, and Most National Brands Barely Register

Sorting every citation by the national brokerage brand behind it shows one company well ahead of the rest.

Compass appears in 86 of the 100 cities and draws more citations than the next two brands combined. Coldwell Banker and RE/MAX follow with real national coverage of their own.

Below that, scale stops helping. Keller Williams is the largest brokerage in the country by agent count and finishes eighth.

Concentrated local presence beats national reach. Shorewest drew more citations than Keller Williams managed nationally, all of them in Milwaukee. Ebby Halliday did the same across the Dallas suburbs.

City By City

No city produced fewer than 18 named agents, and none produced more than 84. This table shows all 100 cities: how many distinct agents or teams any AI platform was willing to name, the single most-cited agent in that market, and how many separate sources AI drew on to get there.

Market size does not predict visibility. Columbus, Ohio produced the second-largest visible field in the country, ahead of Chicago, New York and Los Angeles. Stockton, Santa Ana and Lexington produced the smallest.

How AI Picks, and Where it Breaks

Portals and Directories Supply Almost Every Answer

Listing portals and directories accounted for the overwhelming majority of citations. Agents' own websites made up roughly one citation in nine. Reddit alone finished in the top ten. A discussion forum sits among the most-cited sources in the study.

The platforms trust different things. ChatGPT leans hardest on portals and directories. Gemini pulls more from agents' own websites than any other platform. Google AI Mode barely uses web pages at all. Most of its answers come straight from Google Business Profile listings.

ChatGPT

Google AI Overviews

Gemini

Google AI Mode

Grok

Zillow Sends AI to a List, not to an Agent

Zillow is the second most-cited source in the study, so AI leans on it heavily. The question is which Zillow pages it pulls.

Almost always, it is the city-wide directory: the page listing every agent in the market, rather than an individual agent's profile. Of Zillow's citations, 86.9% point at that city list and only 13.1% at a named agent.

The practical effect is that AI treats Zillow as a place to find agents in general, not as evidence that any particular agent is good. An agent's Zillow profile is doing far less work than the raw citation count suggests.

The Five Platforms Almost Never Agree

The study found no shared consensus on who the best agents are.

Seventy-two agents nationwide were named by every platform that links to agent websites. Even the two Google products disagree with each other.

For an agent, visibility on one platform predicts almost nothing about visibility on another. There is no single AI ranking to win.

Methodology

Local Falcon searched the 100 largest US cities in August 2026 using three phrases: "best realtor near me", "best real estate agent near me", and "top rated real estate agent near me". Each phrase ran on five AI platforms: Google AI Overviews, Google AI Mode, Gemini, ChatGPT and Grok. Every search was repeated across 25 points on a grid covering a 2 mile radius from each city center. That produced 37,500 individual AI queries.

An agent counted as named when AI linked to their own website or to a profile page carrying their name. Sources were ranked by how many of the 25 grid points cited them, so one-off mentions do not outweigh consistent ones. Portals, franchise brands, news outlets, software firms and national services were excluded from the agent count. Sales rankings come from RealTrends Verified 2026, matched by website address.