Who AI Recommends: 123 Local Markets Compared

Who AI recommends for a local service depends on which AI you ask. Across 123 trade-and-city markets checked by BaaDigi (July 28 to October 27, 2026) ChatGPT, Perplexity and Google AI Overviews named 2,578 businesses in total. Only 42 of them (1.6%) were named by all three engines. 2,078 (80.6%) were named by one engine and by no other. Most business-market rows in this cache were associated with only one engine. Only 108 of the 123 markets have all three engines present, and Google AI Overview extraction was incomplete, so missing extraction also affects the overlap shares.
By Ryan Goering, founder of BaaDigi. Last updated: November 2026.
Ryan's take: ChatGPT, Perplexity and Google each pull from different places to build their answer. They're not reading the same sources. So if you're not in the directories and sources one engine leans on, that engine won't cite you — even while another one does. That's why only 1.6% show up in all three. If you want all three sending leads, you have to be present where each one looks.
This post comes from the lookups people ran with our free AI search visibility checker. Each lookup asks three AI engines who they recommend for one trade in one city and stores the businesses each engine named. We pulled the current state of every market in that cache and counted overlap. The method and its limits are spelled out below, and the index of everything we publish is at BaaDigi data and studies.
Can an AI agent actually use your website?
AI agents are starting to visit sites on a customer's behalf — comparing companies and requesting quotes. We run the same agentic browsing checks Google built into Chrome and show you what an agent hits when it lands on your site. No email to run it.
How often do the three AI engines name the same local business?
| Businesses named by | Count | Share of 2,578 |
|---|---|---|
| All three engines | 42 | 1.6% |
| Exactly two engines | 458 | 17.8% |
| One engine only | 2,078 | 80.6% |
Source: BaaDigi AI visibility board cache, 123 markets, 80 distinct trades, 104 distinct cities, checked July 28 to October 27, 2026; 108 of the 123 markets have all three engines present. A "business" is one name on one market's merged list; the same company in two cities counts twice.
How to read the table: the three rows are exclusive and add up to the full list (42 + 458 + 2,078 = 2,578). A business named by all three engines appears once, in the first row. Because matching is by name, a company that one engine lists as "Smith Plumbing" and another as "Smith Plumbing LLC" can land in the one-engine row twice, which inflates the one-engine share by an amount we cannot measure. We did not split the table by the 108 markets that have all three engines present, so we publish no separate shares for that subset.
In 80.6% of business rows (2,078 of 2,578), only one engine named the business. That is the whole finding. It is a business-level companion to the domain-level count in our August study, We Ran 1,429 AI Searches. The Engines Disagree. What changed between the two is the unit: in August we counted which websites the engines cited; here we count which businesses they named. Of the 2,578 business rows, 500 (42 + 458, or 19.4%) were named by two or more engines.
How many businesses does each engine name per market?
| Engine | Markets with a result | Average businesses named | Fewest | Most |
|---|---|---|---|---|
| ChatGPT | 123 | 11.8 | 2 | 32 |
| Perplexity | 110 | 13.3 | 3 | 30 |
ChatGPT averaged 11.8 businesses named per market across 123 markets, with a range from 2 to 32. Perplexity averaged 13.3 across 110 markets, with a range from 3 to 30. The two averages cover different sets of markets, so they are not paired counts. Google AI Overviews is left out of this table on purpose. It returned a result in 121 markets, but the number of businesses our parser captured from it averaged 1.7 with a minimum of 0, which looks like an extraction gap on our side rather than a real engine behaviour. We will not publish "Google names fewer businesses" until we have fixed that, and you should be suspicious of anyone who does on similar data.
Why do AI engines recommend different businesses?
Our data does not say why, and we are not going to guess at the engines' internals. These records do not explain why any engine selected any business. One field is worth reporting as it stands: for 1,588 of the 2,578 business rows (61.6%), the cache contains no domain. This does not identify the sources the engines used, or why the domain is missing.
How did we measure this?
Source: BaaDigi's AI visibility board cache, the table behind the free checker on this site. Each row is one trade-and-city market and holds a merged list of every business any of the three engines (ChatGPT, Perplexity, Google AI Overviews) named for that market, with a flag for which engines named it. Window: markets checked between July 28 and October 27, 2026. Sample: 123 markets, 80 trades, 104 cities, 108 markets with all three engines present, 2,578 business rows. Business-name matching across engines is string-based, so a spelling variant counts as a separate business; this inflates the "one engine only" share by an amount we cannot measure. BaaDigi wrote the lookup prompt template; the trades and cities were typed by whoever used the tool.
What this can support: "most businesses named by one engine are not named by the others," with the 1.6% / 17.8% / 80.6% split, What it cannot support: any ranking claim, including which business an engine placed first. The cache does not keep each engine's ordered list, only which engines named each business, so "position of business X in engine Y" and any rank correlation are not measurable from it. Only 108 of the 123 markets have all three engines present, and Google AI Overview counts are incomplete, so missing extraction also affects the overlap shares. This is a convenience sample of whatever tool users typed, not a designed sample of US local markets, and it says nothing about search volume, clicks or leads.
Who should act on this, and how?
- If you have checked one AI engine and liked what you saw: check the other two before you tell anyone you are "recommended by AI." In this data, 500 of 2,578 business rows (42 + 458, or 19.4%) were named by two or more engines.
- If no engine names you: start with your own site. The cache has no domain for 1,588 of 2,578 business rows (61.6%); that does not identify the sources the engines used. A first-party page per service with a price on it, and a Google Business Profile that matches it, give every engine something to read.
- If one engine names you and the others do not: that is the normal case in this data, not a problem to panic over. Our AI SEO service page explains how we work on it.
- If someone is selling you a single "AI ranking" number: do not buy it yet. Ordered positions are not reliably recoverable from engine output, and ours are not stored for that reason.
Who does AI recommend, and which sources does it cite?
More people ask ChatGPT and AI search “who's the best contractor near me?” every day. See if AI names you — or sends the job to someone else. No email to run it.
Free website audit — see what's costing you leads
No email required. Takes 2 minutes. Completely free.

Ryan Goering
CEO & Founder, BaaDigi
U.S. military veteran and digital marketing strategist who built BaaDigi to help contractors generate predictable leads and revenue. 15+ years in SEO, PPC, and AI-powered marketing automation.
Frequently Asked Questions
Do ChatGPT, Perplexity and Google AI Overviews recommend the same local businesses?▼
Mostly not, at the business level. Across 123 trade-and-city markets checked by BaaDigi (July 28 to October 27, 2026) the three engines named 2,578 businesses. 42 (1.6%) were named by all three, 458 (17.8%) by two, and 2,078 (80.6%) by only one engine. Only 108 of the 123 markets have all three engines present.
How many businesses does ChatGPT name when you ask for a local recommendation?▼
In BaaDigi's board data ChatGPT named an average of 11.8 businesses per market across 123 markets, with a range from 2 to 32. Perplexity averaged 13.3 across 110 markets, ranging from 3 to 30. Google AI Overview counts were excluded because our extraction from it captured an average of 1.7 businesses with a minimum of 0, which looks like a parsing gap rather than real engine behaviour. The two averages cover different sets of markets (123 and 110), so they are not paired counts.
Why do 61.6% of the named businesses have no website recorded?▼
We cannot tell from this data. In the cache, 1,588 of 2,578 business rows (61.6%) have no domain recorded; this does not identify the sources the engines used or why a domain is missing. Publishing your own service pages with real prices gives every engine a first-party page to read.
Is this a representative sample of US local markets?▼
No. The markets are whatever trades and cities people typed into BaaDigi's free checker. It supports the overlap finding, not ranking claims or search volume.
Popular Contractor Marketing Guides
HVAC SEO by City
We serve contractors nationwide — based in Orange County, CA
Explore More Topics
Ready to Get More Leads?
Get a free marketing audit for your contracting business. We'll show you exactly where you're losing leads.
Get Your Free AuditRelated Articles
What Do AI SEO Services Cost in 2026?
Five of 18 agencies we checked publish an AI SEO price. See every published rate, the terms behind it, what software costs, and where our own engines start.
Which Trades Do AI Engines Recommend Agencies For?
We tracked 13,113 ChatGPT and Perplexity answers across 310 BaaDigi-written prompts. For roofing, HVAC, painting and plumbing the most-cited source was Reddit, not an agency.
AI Visibility Index: Oct 2026 Citation Rates
BaaDigi's monthly AI Visibility Index for October 2026: citation rates by engine, the domains AI search cites most, and what changed from the month before.