Your status is the band your share of answers falls in.
Your businessAverage of the 10 regulars: the businesses ChatGPT names most often
Share of voice is not drawn as a second line: it moves with visibility. Point at a report to read its figures.
Your businessAverage of the 10 regulars: the businesses ChatGPT names most often
Point at a report to see how many of its answers showed this business as a card.
Click a metric to see every report. The shaded parts of each chart mark where OpenAI changed the engine under this market.
One ChatGPT answer has two parts, and a business can appear in either: the text ChatGPT writes, and the block of business cards it shows underneath that text. The cards are part of the answer, and each one is built from a directory listing. The diagram in How this was measured shows the anatomy.
The rating your customer reads beside the answer is the one on that Yelp listing: 3.4★ from 25 reviews. Whatever your Google profile says, it is not what they see here.
The top of the market, the businesses ranked either side of your business, and your business. Visibility is the share of the same 30 answers that mention a business, in the text or as a map card. Share of voice is that business's slice of all 341 mentions in the market.
| # | Business | Visibility | Share of voice | Written answer | Places block |
|---|---|---|---|---|---|
| 1 | Wake Plumbing | 100% | 8.8% | ||
| 2 | Plumb Tite Plumbing & Drains | 100% | 8.8% | ||
| ⋯ 2 more businesses between | |||||
| 5 | Preston & Sons | 87% | 7.6% | ||
| 6 | Roberson Brothers Plumbing | 77% | 6.7% | ||
| 7 | Anderson Hills Plumbing INC (example client) | 73% | 6.5% | ||
| 8 | Ken Neyer Plumbing INC | 73% | 6.5% | ||
| 9 | Calla Plumbing | 67% | 5.9% | ||
| ⋯ 25 more businesses below | |||||
ChatGPT named 34 businesses across these 30 answers. You sit 7th.
| What an answer looks like | A Yelp or Google listing builds most of the cards in this market, but they are not the only sources: other directories, and pages from the answer's own web search, build cards too. Whichever listing built a card is where its rating and review count come from. |
|---|---|
| Where we asked | The real ChatGPT web app, asked the way a customer in your city would ask it, as a signed out visitor. Not a developer API. |
| How many times | 30 independent answers, spread across natural phrasings a real customer would use. A single answer proves nothing, because the model gives a different one every time. |
| What counts as a mention | Each answer is read on 2 surfaces: the written text and the places carousel, the block of business cards ChatGPT shows beside its answer. An answer counts once if either happened. |
| How your status is set | From the share of the 30 answers, against the five bands above. The 90% intervals say how tight that share is and never move a status by themselves. |
| How businesses are matched | By map listing id, never by name text, so a business with a similar name in another town cannot be counted as this one. |
| What the charts compare against | The 10 businesses ChatGPT names most often in this market, not every name it happened to mention. A market carries a long tail of names that come up once or twice, and averaging those in would flatter every business measured against them. The 10 are recomputed for every report, and a chain counts once, exactly as it does in the table above. |
| Measurement environment | Engine
gpt-5-6 in all 30 answers, map block present in
30 of 30, 10.8 cards per answer, 11.4
businesses named per answer. The
previous report ran on gpt-5-5. A change of engine moves every business
at once, so the reports either side of it are not strictly comparable. |
| What this does not tell you | Not how many people asked, and not what any one person saw. ChatGPT varies its answers by design and adapts to a person's own history, so these figures describe how often a business comes up for a stranger asking fresh, which is the only thing that can be measured the same way twice. |