SOCi's 2026 Local Visibility Index, covered by Search Engine Land on January 28, 2026, looked at nearly 350,000 locations across 2,751 multi-location brands and found recommendation coverage of 1.2% for ChatGPT, 11% for Gemini, and 7.4% for Perplexity. The same brands appeared in Google's local 3-pack 35.9% of the time. Search Engine Land summarized the size of the problem in one line:
AI visibility is three to 30 times harder to achieve than ranking well in traditional local search, SOCi estimated.
Last verified: September 2, 2026.
Important
Read these percentages as coverage within SOCi's query set, brand sample, and methodology. AI answers can return one business, several businesses, or no local recommendation at all.
They do not mean ChatGPT can only recommend 1.2% of local businesses, and they say nothing about the ranking mechanism behind any single answer. If you need to benchmark one location before building a full scorecard, start with What Is a Good AI Visibility Score for a Local Business?.
The gap is real, and it is smaller once you track the right prompts
A second number is worth putting next to the first. In the Cheers panel for the 28 days ending September 2, 2026, 13 home-services companies, leaving out questions that name the company itself, aggregate only, the median company appeared in 29.5% of its tracked AI answers and all 13 were named at least once. That is not a contradiction of SOCi. It is a different question. SOCi measured a broad brand sample against a fixed query set; the panel measures companies that already chose to track 593 buying prompts between them in their own markets and act on the misses.
The lesson is not that one number is wrong. It is that a coverage rate is only meaningful with its prompt set attached. Mean company appearance inside the panel still varies by engine: 32.8% on Google AI Mode, 27.3% on ChatGPT, 23.7% on Gemini, and 16.2% on Perplexity, a 16.6-point spread, so even a well-tracked brand should expect different results by engine. Cheers does not publish a trade-level rate from this panel because no trade has ten companies behind it. The cuts are in AI visibility statistics and the sampling rules in the methodology.
Cheers panel, by engine
The gap moves by engine
Mean company appearance rate by engine, Cheers panel, 28 days ending September 2, 2026
13 companies
Named companies least often
Cheers panel, 28 days ending September 2, 2026: 13 home-services companies, each measured on all four engines and counted once, non-branded prompts, aggregate only, corrected September 28, 2026. Trade-level rates are not published because no trade has ten companies behind it.
| Engine | Mean appearance rate |
|---|---|
| Google AI Mode | 32.8% (13 companies) |
| ChatGPT | 27.3% |
| Gemini | 23.7% |
| Perplexity | 16.2% (Named companies least often) |
The shortlist math
Google shows you options. Here are three plumbers. Pick one. The user makes the decision.
AI answers often compress the visible shortlist, although the number of businesses varies by prompt and product. The customer may still verify sources, read reviews, visit websites, or run another search.
This changes what teams need to measure. A local 3-pack displays three businesses, while an AI answer may name one business, several businesses, or none. Visibility in one format does not guarantee a mention in the other. Multi-location teams should measure that by market, prompt, and source, using a workflow like How to Audit AI Search Visibility Across Locations.
And these are large surfaces. Sam Altman put ChatGPT at 800 million weekly active users at OpenAI DevDay on October 6, 2025, roughly double the count reported in early 2025. Google's I/O 2026 Search post from May 19, 2026 said AI Mode passed one billion monthly users with queries more than doubling every quarter since launch.
These products are large enough to measure alongside traditional local search.
What the study does and does not explain
The study measures which brands appeared. It does not publish the internal decision rules used by ChatGPT, Gemini, or Perplexity. Operators can still audit four observable areas:
Cited sources. Record the exact pages used in the answer and whether they accurately describe the business.
Public business facts. Correct wrong phone numbers, addresses, hours, service areas, and branch relationships across the sources customers and crawlers can reach.
Customer evidence. Maintain a policy-compliant review program and useful proof of real work. Do not treat review count, wording, or recency as a known AI ranking factor.
Prompt coverage. Test the services and markets that matter, because one brand-level prompt can hide branch-level misses.
What to compare among businesses that appear
Do not infer causation from the winners alone. Compare the businesses that appear with those that do not across observable artifacts, then verify each suspected gap.
Review coverage. Compare legitimate review presence by location and platform, including whether the profile reflects current services. Treat correlations as audit leads, not proof of a ranking factor.
Location clarity. Check whether the branch page, profiles, and applicable structured data agree on the location, services, and contact path.
Fact accuracy. Prioritize wrong or outdated facts over harmless formatting differences.
Independent sources. Record whether relevant local reporting, trade associations, directories, or community discussions appear in the cited results. Do not manufacture mentions.
Useful media. Photos and videos can help customers inspect real work. If YouTube or another media source appears in the answer, document it; do not assume adding video will cause a recommendation.
Pro Tip
Treat a citation as evidence about the source used for that answer, not as proof of a ranking factor or a guaranteed click.
The earned media factor
This is the part most businesses miss completely.
Owned-page and listing cleanup matter, but third-party pages also appear in some answers. The weight of those sources varies and is not publicly documented.
Independent sources such as Reddit, YouTube, news outlets, industry publications, and review platforms may appear in citations. In December 2025, Yext put the split at 86%:
86% of sources in AI-generated answers are from brand-managed properties.
Brand-managed includes third-party profiles a business maintains but does not own, so that number is an argument for keeping records accurate rather than proof that owned content wins.
The Cheers market baseline puts a ceiling on the rest. Across 356,019 home-services citations in 100 metro areas, social and forum domains took 4.3% of ChatGPT citations, 4.8% of Google AI Mode citations, 1.4% of Perplexity citations, and 0.3% of Gemini citations. Reddit was the sixth most cited platform domain overall, present in all 100 metros. Treat genuine customer discussion as public feedback worth reading, not as a channel to seed recommendations.
Independent mentions cannot be purchased in the same way as an owned profile update. Treat them as earned evidence, not as an optimization shortcut.
Next step
Is AI recommending your business?
Find out how visible you are across ChatGPT, Gemini, Perplexity, and AI Overviews.
The multimodal edge
Businesses that rely on text-only web presence are leaving visibility on the table.
Photos, videos, detailed service pages, and applicable structured data can make a site more useful to customers and easier to inspect. Public evidence does not show that adding each format automatically raises recommendation confidence.
Use media when it documents real work, explains a service, or answers a buyer's question. Do not create it only to satisfy an assumed ranking signal.
What the study means for multi-location teams
Do not multiply the study's 1.2% coverage rate across your own location count. The useful next step is a location-level baseline using the services, markets, prompts, and providers that matter to your business.
Dallas and Houston can produce different answers because they have different branch pages, profiles, reviews, competitors, and service facts. Measure each market instead of relying only on the parent-brand result.
If 40 of your 100 locations have conflicting public facts, incomplete service pages, or profiles that no longer match the real branch, that is 40 separate audit targets. It is not proof that an AI product will exclude those locations, but it is enough reason to correct customer-facing errors and retest the relevant prompts.
Centralized ownership helps keep those public facts and workflows current. Hello Sugar used a centralized review process across its franchise network, but that review growth should not be presented as proof of a universal AI recommendation effect.
A 90-day audit and cleanup plan
Days 1 to 30, foundation. Audit every location's review profile: count, recency, and what the reviews actually say. Run a consistency check across Google, Yelp, BBB, and the industry directories that appear in your own answers, and write down every conflict rather than fixing them ad hoc. Put LocalBusiness markup on every location page with the specific subtype, Plumber or BeautySalon rather than the generic parent. Then build the baseline: ask ChatGPT and Gemini your category question in every market you serve and save what comes back. The Cheers AI Visibility Grader does a focused one-profile version of that across ChatGPT, Gemini, and Perplexity in a few minutes.
Days 31 to 60, velocity. Start a review collection program you can sustain, measured as new reviews per location per month, with compliant asks and enough detail in the review to be useful to a reader. Fix material source errors before anything else: wrong phone numbers, old addresses, conflicting hours, services you no longer sell, duplicate profiles. Add photos or video where they document real local work. Rewrite thin or inaccurate service and location pages, and resist the urge to spin up a page per keyword variation.
Days 61 to 90, amplification. Give accurate information to local news, industry publications, associations, and communities when there is a real story or resource behind it. Add schema only where it matches visible content and the page's actual entity relationships. Re-test the same prompts against your day 1 baseline, record which source and which answer changed, and hold off on attributing movement to a single tactic until it repeats.
Pro Tip
Track your AI visibility monthly. Ask the same questions across ChatGPT, Gemini, and Perplexity for every market you serve. This becomes your GEO scorecard.
Sources
Every link below was opened and checked on September 2, 2026.
- AI local visibility is up to 30x harder than ranking in Google. Search Engine Land, January 2026, covering SOCi's 2026 Local Visibility Index. Source of the 1.2%, 11%, 7.4%, and 35.9% coverage rates across nearly 350,000 locations.
- 15 AI search stats every marketer needs to know going into 2026. Yext, December 2025. Source of the 86% brand-managed properties figure.
- ChatGPT statistics you should know for local SEO. Local Falcon, November 2025. Share of prompts that trigger a live search and the local-intent share of those searches.
- Sam Altman says ChatGPT has hit 800M weekly active users. TechCrunch, October 2025.
- A new era for AI in Search. Google, May 2026. AI Mode monthly users and query growth.
- Optimizing your website for generative AI features on Google Search. Google Search Central, last updated July 2026.
- Cheers panel and market baseline, 28 days ending September 2, 2026. 13 home-services companies and non-branded prompts for appearance rates; 32 prompts across 100 metro areas and 356,019 citations for the source mix. Aggregate only, corrected September 28, 2026. Methodology.
Dylan Allen-Arnegård is the CEO of Cheers, the local search platform for service businesses.