How often does the average local business get cited by AI?

Not published yet. The mean citation rate will be published with its sample size and confidence interval once the sample clears our minimum of 25 businesses.

How was AI visibility measured?

For each business, category-level buying-intent queries were run against three web-grounded AI assistants and sampled repeatedly, because these systems return different answers to the same question. Citation rate is the share of sampled answer-slots in which the business was named. Failed measurements were discarded rather than recorded as zeros.

Original research · ASN Intelligence

We are measuring local businesses in AI search. The rate publishes when the sample clears 25.

Q: What percentage of local businesses are invisible in AI search?

ASN Intelligence is measuring this now. We publish a rate only once the clean sample reaches 25 businesses; as of September 2026 it has not, so no percentage is published yet. The method is on this page.

Every AI-visibility company will tell you that businesses are missing from ChatGPT answers. Almost none of them publish a number, a sample, or a method. Here is ours, including the parts that weaken it.

Published 5 September 2026 · revised 7 September · Sample: 36 businesses · Engines: ChatGPT, Gemini, Perplexity

—share never named in any sampled AI answer for their own category: withheld until the clean sample reaches 25 businesses

—mean citation rate: withheld until the clean sample reaches 25 businesses

36distinct local service businesses, one data point each

The headline finding

We are measuring local service businesses across ChatGPT, Gemini and Perplexity on buying-intent questions for their own category. We publish a rate only when the clean sample reaches 25 businesses; as of September 2026 it has not, so the headline figures on this page are withheld rather than rounded up from a smaller sample. The method below is exactly what will produce them.

The businesses that were visible were not reliably visible either. The mean citation rate across

The mean citation rate and the share of businesses never named will be published here with their sample size and confidence interval once the sample clears 25.

How it was measured

Queries. For each business, a set of category-level, buying-intent questions in its own city or service area. Not brand-name lookups — asking an assistant about a business by name proves nothing, because it will find almost anyone that way.
Engines. Three web-grounded assistants: ChatGPT, Gemini and Perplexity.
Repeat sampling. Every query was run multiple times. These systems return different answers to the same question, so a single run is not a measurement.
Citation rate is the share of sampled answer-slots in which the business was named. A business named in 9 of 28 sampled slots scores 32.1%.
Failed runs were discarded, not scored. When a query errored or returned nothing, it was dropped. It was never recorded as a zero. A zero has to be an observed absence, not a failed observation.
One data point per business. Where a business had several tracked query sets, only its most recent score counts, so no single business can weight the average twice.

What the sample is

Attribute	Detail
Size	36 businesses
Type	Local service businesses — deathcare, land clearing and forestry mulching, tree and landscaping services
Geography	Predominantly Virginia (Northern Virginia, Charlottesville and surrounding counties)
Period	Measurements taken between July and September 2026
Selection	Businesses appearing in the competitive set for categories ASN tracks; not a random national sample

What this study does not show. 36 businesses is a small sample concentrated in two industries and one state, so the exact percentages should not be read as a national figure for all local businesses. The selection is not random. And AI answers move — a business invisible in September may be cited in October, which is the entire point of measuring repeatedly rather than once. What the sample does support is the direction and the rough magnitude: being absent from AI answers in your own category is common, not exceptional.

Why it matters

Traditional search gives a business ten blue links to compete for. A generative answer typically names between two and seven businesses, and the rest are simply not mentioned. There is no page two. A business that is not in that list is not losing rank — it is absent from the conversation, and the buyer never learns it exists.

The most common reason is not the website. Assistants assemble answers from the sources they already trust: directories, trade associations, review platforms and roundup articles. A business with an excellent website and no presence on any of those sources will stay invisible no matter how good the site is.

Methodology notes and reuse

The figures above come from live measurement data rather than being written by hand, but this page itself is updated manually — deliberately, so the numbers a reader or an AI assistant sees are static and citable rather than shifting under them mid-sentence. Each revision is dated at the top. The underlying method is the same one ASN runs for clients every day.

Journalists, analysts and other agencies are welcome to cite these figures with attribution to ASN Intelligence LLC and a link to this page. If you want the per-industry breakdown or want us to run the same measurement on a sample you choose, get in touch — we would rather the number be checked than repeated.