Your phone rings from two places now: paid engines like Google LSA and Meta Ads, which you already measure to the dollar, and a third engine you may not measure at all. AI answers. That shift matters whether you serve Monmouth County or customers elsewhere. One AI check tells you almost nothing: the same prompt returns answer sets with only 20–33% overlap on average. If you don’t measure your mention rate over time, you’re flying blind on a channel that 45% of consumers say they use for local business recommendations.
AI visibility is not a position to check once and file away. AI answers change. You need a repeatable system that measures how often you appear, whether you keep appearing, and which competitors share the answer.
Why one AI search tells you nothing
BrightLocal’s 2026 study analyzed more than 200,000 non-branded local prompts, across 1,342 locations, three AI platforms, and multiple runs over 60 days. It examined nearly 1.97 million citations. The results show why a one-off search is a poor performance report.
- Repeat answers overlap by 23% on ChatGPT, 21% on Google AI Mode, and 33% on Google AI Overviews.
- A business that appears once reappears in about half of later attempts: 50% for ChatGPT and AI Mode, 58% for AI Overviews.
- 71–80% of businesses appear in half of attempts or fewer.
- Google Maps mentions a tracked business in 66% of searches. AI surfaces mention it in 32–38%.
- Maps shows 99% agreement across attempts, compared with 91–96% for AI platforms.
Variance is the amount the list of businesses changes between runs. Persistence is how often a business returns after appearing. They measure different things. An answer can change substantially while a particular business still reappears in roughly half of checks.
That means AI visibility is a percentage game, not a yes-or-no result. Measure the rate, not the result of one search. See the BrightLocal local AI visibility study for its methodology and findings.
The three numbers that actually matter
Track metrics that tell you whether your visibility is becoming more dependable and competitive.
1. Mention rate
Mention rate = checks where your business is named ÷ total checks.
Calculate it by platform and prompt group, not just as one blended score. BrightLocal’s 32–38% AI mention rates are a useful reality check against the 66% Google Maps result. They are study averages, not a promise or target for your business. Report raw counts beside percentages (for example, 6 of 20 checks (30%)) so a small sample does not disguise the scale of the result. Record empty AI responses separately.
2. Persistence
Persistence shows how often your business appears again in later runs after it has appeared once. The study’s baseline was roughly 50% or higher, depending on the platform. Use this metric to distinguish a one-time appearance from a presence that recurs across your measurement window.
3. Share of voice
Share of voice shows how your mentions compare with the competitors named in the same answers. Track which businesses keep appearing alongside yours. A competitor’s rising share may point to a gap worth investigating; it does not prove the cause by itself.
Keep a monthly log with:
- Prompt and platform
- Date and search location
- Whether you were named
- Competitors named
- Sources or URLs cited
- Empty response, if applicable
Keep the raw counts and percentages together. That simple discipline turns a collection of screenshots into an AI visibility tracking record you can compare over time.
How to run the checks without wasting your week
Build a fixed panel of 10–20 non-branded prompts that reflect how customers actually ask for help. Include emergency, price, best-of, and comparison questions. For example: “need an emergency plumber open now in Freehold.”
Then follow a consistent protocol:
- Use the same prompts every month. That keeps your prompt cohort comparable.
- Check each platform separately. Run the prompts in fresh, signed-out sessions to reduce personalization effects.
- Repeat checks over a multi-week window. A daily swing can be noise; repeat result sets overlap only 20–33% on average.
- Log the answer, not just the mention. Save the businesses named, the sources cited, and whether the response recommends you for the relevant service.
- Keep locations consistent. Where service geography matters, record the search location along with the prompt.
Platform choice matters. ChatGPT returns an average of 4.1 businesses per response (median 4); AI Overviews returns 2.5 on average (median 3). Empty responses occur in 10% of ChatGPT checks and 11% of checks on each Google AI surface. The platforms also disagree: AI Mode and AI Overviews share 29.3% of named businesses, while ChatGPT overlaps just 19–20% with either. They are not interchangeable, so check each one you care about.
You can track this manually in a spreadsheet. Tools are available, too: BrightLocal’s Local AI Visibility Tracker reports a 0–100 score per location, supports up to 20 prompts, and checks each location twice a month. It reports mentions, sources, share of voice, prompts, and competitor performance. Use a tool if it saves time, but keep the measurement rules visible and the same from one period to the next. BrightLocal describes its tracker in its launch announcement.
What the numbers tell you to fix
Your tracking system should help you decide what to improve, not just show a dashboard.
Start with geography. BrightLocal found that 52.8–59.3% of AI selections were within 5 km of the search point and 71.7–78.2% were within 10 km. ChatGPT has a longer tail: its 90% radius stretched to 287 km, compared with 64 km for AI Mode and 38 km for AI Overviews. Make your address, service area, and the places you serve easy to understand. Keep your name, address, and phone number consistent across your site and listings.
Make your website useful to people and AI. Of roughly 116,000 source domains, about 93% were business websites; those sites accounted for 42% of all citations. Put clear service details, location information, relevant expertise, and customer reviews on your site. Make service and location pages straightforward to navigate and easy to interpret.
Check platform-specific sources. Google Business Profile represented 28.5% of citations in the study and is used heavily by Google’s AI surfaces, but hardly used by ChatGPT. Yelp was important across platforms; ChatGPT also leaned on Yelp and Bing, while Google’s AI platforms drew more heavily on Google Business Profile and Facebook. Audit the sources you see in your own checks instead of assuming every platform relies on the same signals.
Account for your business model. Storefronts had a 33.6% mention rate, compared with 29.4% for service-area businesses. If customers do not visit your location, make your service-area coverage especially clear across your website and business profiles.
Close the loop with the channels you already measure
AI visibility is an early signal in your growth engine. The phone call is still a conversion event. Your Google LSA campaign can capture high-intent calls; Meta Ads can generate lead forms and call campaigns; AI recommendations can shape discovery before a prospect contacts you.
When your AI mention rate rises, watch branded search, direct calls, and qualified leads. Treat those changes as signals to investigate, not automatic proof that AI caused the lift. Use call tracking and your lead records to connect visibility trends with calls, appointments, and revenue. That is how you measure the engine end to end.
Build a measurement system you can act on
Don’t wait for a perfect attribution model before you start measuring. Set your prompt panel, record the same fields every month, and compare multi-week trends by platform. Then use what you learn to prioritize website clarity, local information, and the sources that actually appear in your results.
Partner with SEOPulseAI for a clearer view of what is working and what to improve. Explore our services and request a free, no-obligation review. Or visit SEOPulseAI to start the conversation.