The data

We do not guess what AI says. We count it.

Every month we ask the AI engines real customer questions about real local businesses, keep every answer, and read all of them. This is what that adds up to, what it shows, and how it is measured.

120,000+

AI answers collected and read

What we tell you about your market is counted, not guessed.

260,000+

sources AI used to decide, logged

We know what the engines trust before we spend a dollar of your money.

150,000+

searches the engines ran, recorded

We see how AI goes looking for a business like yours.

80,000+

visits from AI engines to our clients

The engines keep coming back to read about the businesses we work for.

12,800+

different websites the engines cited

400+

competitors tracked by name

19

different AI crawlers identified

Across ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews. Rounded down.

Finding 1 · The mechanism

AI recommends what it has read.

When an engine had read one of the business's own pages before answering, it named that business 84% of the time. When it had not, 16%. The gap holds inside every engine on its own.

Read a page of theirs Did not
Gemini
92.2%
20.7%
Google AI Overviews
89.0%
33.4%
Claude
86.6%
11.9%
Perplexity
81.3%
16.7%
ChatGPT
79.4%
21.1%

Share of AI answers that name the business. 64,441 answers, 2026. This compares answers to each other, so it is not affected by which questions were asked or by model updates.

Finding 2 · The starting line

Seven in ten AI answers recommend someone else.

That is what a first scan shows, before any work is done. Most local businesses have never seen this number for themselves.

70%

of answers named a competitor, or nobody.

30%

named the business being measured.

18,613 answers from first scans, before any work had been done.

Finding 3 · The readers

AI crawlers show up once there is something to read.

80,000+ visits from 19 different AI crawlers to the pages we publish for clients. These come from server logs, not from scans.

Visits per month

June
1,092
July
17,476
August
42,067

By crawler, all time

ClaudeBot
17,471
GPTBot
13,991
Amazonbot
10,389
OAI-SearchBot
8,861
ChatGPT-User
7,582
Meta AI
5,120
PerplexityBot
4,160

Only crawlers that identify themselves are counted.

Finding 4 · The results

Our clients win the answer.

36%

more AI mentions than their strongest local rival

When a customer asks AI who to call, your name comes up before the competitor down the road.

95%

of the competitors AI names trail our client

You stop sharing the answer with every other business in town.

49%

of mentions are the first recommendation

Being on the list is not the goal. First pick is the one the customer calls.

99.5%

of mentions are positive or neutral

AI does not just bring you up. It speaks well of you when it does.

Totals across every location we manage, on each one's latest scan, comparing our client and its competitors in the same AI answers. Locations still behind their rival are included in every figure.

Read the case studies

Original local data

How much is rent in Jackson, WY?

The Jackson Apartment Rent Index: asking rents at every named apartment complex in Jackson, by unit type, market-rate and affordable, with pets, parking and the source for every number. Updated monthly.

See the October 2026 index

Method

How we count.

Real answers, kept in full.

Every answer an engine gives is stored word for word. Nothing is sampled after the fact.

Every answer is read the same way.

Was the business named, was it the first pick, and who else was named. Only answers that have been read this way are counted in any percentage here.

Like for like, or not at all.

A before and after only counts when the question wording and the engine are identical both times. If the wording changed, we report a standing, not a change.

Broken instruments are excluded.

In July 2026 two engines returned nothing for every business. Counting their recovery would have shown a gain everywhere, so they are left out of that comparison.

Margins of error are shown.

AI answers vary from run to run. Movement inside the margin is reported as steady.

The misses stay in.

Locations that are still behind their rival are counted in every total on this page.

What we do not publish.

How we move the numbers. That is the part clients pay for, and we are glad to walk you through it on a call.

See how this would work for your business.