Research · recurring benchmark
Real Estate AI Visibility Index
A recurring, market-by-market benchmark of how often AI recommends real estate teams — built so each edition can be compared with the last.
- Status
- First cross-market edition in progress
- Provider
- ChatGPT (OpenAI)
- Method version
- v1.0
- Last updated
- September 2026
What the index measures
- Recommendation frequency
- How often AI recommends each observed team across a locked question set.
- Production vs visibility
- Differences between verified closed production and measured recommendation frequency.
- Sources around recommendations
- Which classes of source appear in the answers where each team is recommended.
- Entity clarity
- How consistently each observed team is described across public sources.
- Market-level patterns
- How concentrated recommendations are within a market.
- Question-level differences
- Which question types produce different winners.
The index is not a universal ranking of real estate teams. It reports what was observed in specific markets, on a specific provider, in a specific window, using a published question set. Where a market has not been measured, the index says nothing about it.
Key findings
The first cross-market edition has not been published. One single-market benchmark is complete and available in full, and it is the only measured data currently reported.
| Observed team | Answers recommending | Share of captured answers |
|---|---|---|
| The team (anonymized) | 4 / 256 | 1.6% |
| Team A | 116 / 256 | 45.3% |
| Team B | 52 / 256 | 20.3% |
| Team C | 48 / 256 | 18.8% |
| Team D | 40 / 256 | 15.6% |
Teams anonymized. Shares may sum above 100% because a single answer can recommend more than one team.
Definitions used by every edition
- Prompt categories
- Hire-intent, best-of, neighborhood, transaction-type and record or event questions. The mix is fixed before capture and published with the edition.
- Sampling
- A locked set of 64 buyer and seller questions per market, written in the language people actually use, not keyword strings.
- Repetition
- Each question asked 4 separate times in independent sessions, because repeated answers differ.
- Recommendation share
- Answers that put a team forward to hire or contact, divided by the valid-answer denominator for that edition.
- Citation share
- Citation instances attributed to a source class, indexed against the most-cited class. Counted per citation, not per answer.
- Valid-answer denominator
- 256 per market edition — 64 questions × 4 repetitions — counting only answers that were captured in full.
- Exclusion rules
- Refusals, truncated captures, and answers that did not address the question are excluded before counting and reported as such.
- Provider and model metadata
- ChatGPT (OpenAI). Exact model metadata is unavailable for these benchmark captures and is not inferred.
Market coverage
| Market | Status | Valid answers |
|---|---|---|
| Greenville, South Carolina · residential | Published (anonymized) | 256 |
| Jersey City, NJ · residential | Published (anonymized) | 256 |
| Additional markets | In progress | Not yet measured |
Markets marked in progress have not been measured. No figure is estimated, modelled or carried across from another market.
Methodology and source notes
| Field | Value |
|---|---|
| Provider | ChatGPT (OpenAI) |
| Question set | Fixed buyer and seller question set per market |
| Repetitions | 4 per question |
| Sample size | 256 answers maximum per market edition |
| Classification | Recommended · mentioned · absent, assessed per answer |
| Anonymization | Anonymized by default |
| Methodology version | v1.0 |
What this cannot prove
- This is not a universal ranking of all real estate teams.
- Results describe the captured window and provider, not AI systems in general.
- Nothing here demonstrates causation between any source, action and outcome.
Citing this report
Cite as: Recommended First, Real Estate AI Visibility Index, September 2026, methodology version v1.0. Please link to this page rather than reproducing tables without their conditions. For the underlying preserved answers, contact francisco@recommendedfirst.com.
Keep reading
See who AI recommends in your market — and where you stand.
One client per market. Start by telling me where you work.