We built an index to measure one thing: when a buyer asks AI who the best real estate agent is, does a real agent show up? We tested 73 top-producing agents on live AI engines. Exactly one appeared.
In a California cohort of 55 agents — every one selling $31–40 million a year — zero surfaced when we asked for the best agent in their own market. In the national pilot we scored end to end, the average AI Visibility Score was 62 out of 100: findable, but not yet recommendable.
I build this for a living, and I ran it on my own company first — LeadJens scored a 5 the first time out. So this is not a gotcha. It's the plain state of the field: doing the job well no longer means AI can see you, and almost nobody has closed that gap yet.
What the Index measures
The AI Visibility Score rates an agent from 0 to 100 across five things AI reads when it decides who to name. Each is worth 20 points:
- Structured data — does the site expose valid
RealEstateAgent/Person/Organizationschema, including reviews? AI reads this first. - Google Business Profile — a verified profile with the right category, complete info, and activity.
- Machine-readable reputation — reviews exposed in a form machines can read and attribute to the agent, not just stars sitting on a third-party site.
- Cross-profile entity consistency — the same name, photo, brokerage, and contact across Zillow, Realtor.com, Homes.com, Yelp, GBP, and LinkedIn. Conflicts split the identity AI needs to resolve.
- AI answer presence — when you actually ask an engine "best realtor in [city]," does the agent appear, and is the description correct?
Scores fall into three bands: 0–39 Invisible (AI can't find or can't trust them), 40–69 Emerging (findable, not yet recommendable), and 70–100 Recommendable (set up to be surfaced and named).
The headline: agents are invisible at the exact moment of the question
Across 12 California metros, we asked AI the question a buyer actually asks — "best real estate agents in [city]" — and 0 of the 55 agents we tracked appeared, even though each sells $31–40M a year. Fold in the national pilot and it's 1 of 73 agents surfacing on the discovery query — roughly 99% invisible. The one who did appear had published a local market blog for two decades.
AI answered the question three ways, and none of them included the agents we tracked:
- It named the usual suspects pulled from directories — Yelp, HomeLight, FastExpert, YouTube round-ups — often mega-teams a solo agent is explicitly not part of.
- It named only brokerages, no people (Menlo Park and Napa returned Compass, Sotheby's, Coldwell Banker, and nothing else).
- It named no one at all — San Diego, Danville, and Sacramento returned only directories, and one market returned literal "Agent A (example), Agent B (example)" placeholders because AI had no real names to give.
A strong reputation isn't a machine-readable one
Only 2 of the 18 fully-scored agents — 11% — exposed their reviews in a format machines can read (an aggregateRating in their own site's schema). Almost every agent had a wall of real five-star reviews; they just lived on third-party sites AI can't reliably credit to the agent. A 5.0 the machine can't attribute to you does nothing for you at the moment of the question.
Most agents' identity is fragmented
Entity fragmentation hit the majority of the sample — stale brokerage tags, name variants, split brand-versus-brokerage identity, and same-surname family pairs the machine can merge by mistake. Structured data was the weakest dimension overall: only about a third of the pilot carried a valid agent-entity in their schema. When AI can't resolve who someone is, it doesn't recommend them — the mechanics of that are in why AI confuses you with another agent.
The average agent is "Emerging," not "Recommendable"
The national pilot averaged 62 out of 100. Here's how the 18 agents split across the bands:
The gap isn't talent — every agent in this sample is a high producer. The web they built their business on was designed for humans. The machine now answering the buyer's question can't read it.
What this means if you're an agent
The five signals the Index measures are the same five you can fix, and they move faster than most agents expect. One client of mine, Andrew, started at 17/100 with AI calling him a San Francisco commercial broker when he's a San Mateo property manager and REALTOR®. After the fixes, a stranger called and said he'd found him through an AI query — 18 days after the assessment. That's one result, not a guarantee; I promise visibility, never the lead itself. You can read the full case study, or the five signals that move AI visibility.
How we ran it
Everything here comes from public sources and a reproducible, dated method. The sample is top-producing solo agents drawn from RealTrends Verified 2025 (2025 production), in the $31–40M-a-year band, across 12 California metros — plus a national pilot of 18 agents across 9 markets. AI answer presence was measured with live Perplexity discovery queries on 2026-06-26; structured data was read from each site's page source; Google Business Profiles, reviews, and cross-platform profiles were checked by hand.
Honest limits (so the numbers stay checkable):
This is a pilot, and it's directional. The full 0–100 score across all five dimensions was computed for the 18-agent national cohort; the California cohort was measured on the discovery dimension (AI answer presence), testing one representative city per metro — 55 of 131 sampled agents. Agents are anonymized. The rubric was built for individual agents; applying it to brands (including LeadJens's own score) is a looser fit. The full study — 150 agents, 10 markets, every dimension scored on live engines — is next, and will tighten these numbers rather than overturn them.
Common questions
What is an AI Visibility Score?
A 0–100 rating of how findable and recommendable an agent is when someone asks AI who to hire, built from five signals: site structured data, Google Business Profile, machine-readable reviews, cross-profile consistency, and whether the agent appears when AI is asked for the best agent in their market.
Why don't top-producing agents show up when you ask AI for the best agent?
Because volume and AI visibility are unrelated. AI answers from what it can read — structured data, machine-readable reviews, consistent profiles. Agents selling $31–40M a year were invisible on the discovery query because their reputations weren't in a form the machine could read, not because they weren't good.
Do good Google reviews help an agent show up in AI?
Only if AI can read and attribute them. In the pilot, just 2 of 18 agents exposed reviews as machine-readable data in their own schema. The rest had strong reviews stuck on third-party sites, where AI can't reliably credit them.
How is AI visibility different from SEO?
SEO ranks a page on a results screen. AI visibility is whether AI names you as a person when someone asks who to hire — optimizing you as an entity AI can identify, trust, and recommend inside its answer.