The Answer Gap Index™ · Edition 1 · September 2026
We asked Google’s AI Overview, Google’s AI Mode, ChatGPT and Perplexity to name the best local business, 720 times. Then we checked every business they named against Google’s map pack and top 10 for the same search.
The Answer Gap Index™ is our ongoing measurement of the gap between who search engines rank and who AI recommends. Edition 1 reports six measures, not a single score.
Free · three minutes · no email. See which of six gaps costs you the most booked jobs, including whether buyers can find you.
Ghost Winners™
60.5%of the businesses AI recommended were in neither Google’s map pack nor its organic top 10.
Ghost WinnersIn the map packTop 10 only
Base: 673 recommended businesses, each checked against five captures of the same city’s search results.
TL;DR
Tap any number for the full finding.
Cite this study
Andrew Ryan Marketing (2026). The Answer Gap Index™, Edition 1: 720 AI answers compared with Google’s local rankings. Data collected 28 September 2026. https://andrewryanmarketing.com/answer-gap-index/
Use it freely. Just link back. The numbers, charts and data are licensed CC BY 4.0: share or adapt them with credit to Andrew Ryan Marketing and a link to this page.
Edition 2 is collected 26–30 October 2026 with the same questions, cities and engines, so every measure can be compared.
For every business an engine recommended, we checked whether it held a spot in Google’s local 3-pack or organic top 10 for the same search — in any of five captures. The orange share is the businesses that held neither. We call them Ghost Winners™.
In the map packTop 10 onlyNeither (Ghost Winner)
The short answer: 58.1% of the local businesses ChatGPT recommended were in neither Google’s map pack nor its organic top 10 for the same search. Across all four engines, 60.5% of the 673 businesses named were Ghost Winners — and the rate held across home services (65.5%), professional services (57.3%) and B2B services (59.3%).
Observed by Andrew Ryan Marketing, 28 September 2026. Unique businesses per city and engine; 60 businesses we could not resolve to a website are excluded, which lowers the Ghost rate rather than raising it.
Free chart for download
The first business an answer names is the one a buyer sees first. We checked how often that first pick held no map-pack or top-10 spot for the same search.
Google’s AI Overview is the exception: its first pick was in the map pack 73% of the time.
Free chart for download
On “best [business] in [city]” searches, the classic AI Overview appeared about a third of the time. Google’s AI Mode answered almost every time.
So the AI answer a local buyer is most likely to get from Google comes from AI Mode, where 53.2% of the businesses named were Ghost Winners.
We asked each engine the same question five times. Each strip below is the 36 city-and-business pairs; a lit square is a pair where the same business came first every time.
Perplexity was measured through its API at default settings, not the consumer app, which likely explains its steadiness. The AI Overview never appeared in all five runs for any pair, so it has no lock rate.
Same city, same question, same moment. We compared ChatGPT, Perplexity and Google AI Mode side by side across the 160 runs where all three named businesses.
Being named by one engine says little about the other two. AI visibility has to be measured engine by engine.
For every recommendation, we checked whether the answer cited the business’s own website, or only other sites — directories, “best of” lists, review platforms and forums.
The short answer: only 7% of ChatGPT’s local recommendations came with a citation to the business’s own site. The proof it leaned on lived somewhere else.
Answer cites the business’s own site
Counted per recommendation, across all runs. Engines don’t tie each citation to a specific business, so this measures whether the answer as a whole cited the business’s own site.
Free chart for download
Follow-up study · Why they get named
We took the businesses Google already showed in its map results for the same 36 searches, then compared the ones AI named with the ones it never named — 25 factors each, with the method fixed before we collected anything.
Made no difference
And the Ghost Winners™? On these signals they looked like the named businesses Google did show. What set them apart was distance: a median 6.8 miles from the city-centre search point, against 1.7. AI recommendations appeared less constrained by distance than Google’s local results — but moving the search point explained only part of the gap (see Method).
Follow-up study, same 36 searches, 28 September 2026. 482 businesses: 246 Google showed and AI named, 117 Google showed and AI never named, 119 strong Ghost Winners. Medians shown. These are correlations: they show what AI-named businesses have in common, not what makes AI name them.
What it means for a local business
A ranking report can’t tell you whether AI names you. The only way to know is to ask the engines your buyers use, the way your buyers ask.
AI leaned on directories, lists, reviews and forums far more than on the business’s own website. What others say about you is the evidence it reads.
The #1 answer changed from run to run and from engine to engine. Being named has to be measured repeatedly, on every engine that matters.
Free · three minutes · no email. Four numbers you already know show which of six gaps costs you the most booked jobs.
Already know your gap? Book Your 20-Minute Gap Review →
Method
The method was fixed and written down on 28 September 2026, before any data was collected. Every Edition 1 number on this page comes from that single collection. The follow-up study section is labelled as such.
Six measures, not a single score. Every edition reports them the same way, so the numbers can be compared over time.
The PDF is a branded two-page summary. The CSV has 144 rows: every city, business type and engine, with runs, businesses named, Ghost Winners and how often the #1 pick repeated. Free to use with a link back to this page.
Questions
Not reliably. In the Answer Gap Index (720 AI answers, September 2026), 58.1% of the businesses ChatGPT recommended were in neither Google’s local 3-pack nor the organic top 10 for the same search. Ranking helps, but it does not decide who AI names.
A business an AI engine recommends while it holds neither a Google local 3-pack spot nor an organic top-10 result for the same search. In Edition 1 of the Answer Gap Index, 60.5% of the 673 AI-recommended businesses we could resolve were Ghost Winners.
In our follow-up study of 482 businesses (September 2026), among businesses Google already showed in its map results, the ones AI named had more Google reviews (median 106 vs 64), more recent reviews and more websites linking to them than the ones it never named. Star rating, schema markup and homepage length made no difference. These are correlations, not proof of cause.
Rarely. Across 160 city-runs, ChatGPT, Perplexity and Google AI Mode named the same business first 6% of the time, and in 52% of runs they had no recommended business in common.
Mostly from other websites. Only 7% of ChatGPT’s recommendations in the study came with a citation to the business’s own site; the rest were supported by third-party sources such as directories, rankings and review sites.
The same prompt, “best [business type] in [city, state]”, was sent to Google AI Overview, Google AI Mode, ChatGPT and Perplexity five times each for 12 business types in Denver, Boise and Chattanooga on September 28, 2026. Named businesses were matched against five captures of Google’s local 3-pack and organic top 10. The method was fixed before data collection.
No form. No sales team. No callback queue. Every channel below goes straight to Andrew.
423-299-1718 · Johnson City, TN · The person you reach is the person accountable for your results.