Key takeaways
- AI Overviews appear for a large share of local queries (studies put it between 43% and 68%, depending on the dataset), and they show up far more for informational and comparison queries than for simple "near me" searches.
- AI Overviews ignore the map pack's proximity logic. A business can rank #2 in the local pack and be completely absent from the AI answer for the same query.
- Checking visibility by city takes a repeatable process: a fixed query list, run per city, logged monthly. Single checks are unreliable because AI Overview citations change roughly every two days.
- Google increasingly cites its own properties (Business Profile, Maps, YouTube) inside AI Overviews, so being complete inside Google's ecosystem matters as much as your website.
- Tools like Local Falcon, Promptwatch, and BrightLocal can automate city-level AI visibility tracking so you're not running hundreds of manual searches every month.
What AI Overviews do to local search
If your local traffic has felt off lately, this is probably why. A Whitespark study found AI Overviews appear for 68% of local business queries, while local packs appear for only 39% of the same queries. That's a 29-point gap where customers get an answer without ever seeing the map results you've spent years protecting.
The trigger rates vary wildly by intent, which is the part most local businesses miss. Search Engine Land's analysis of local queries found AI Overviews show for just 15% of simple local-intent searches ("tacos san francisco") but 92% of informational ones ("how long does an eye exam take near me") and 97% of hybrid commercial queries ("average cost of dental implants in phoenix"). The local pack still owns the simple stuff. The AI owns everything with a question or comparison baked into it.
Here's the part that changes your strategy: AI Overviews don't work like the map pack. Local Falcon's research across 4,400+ businesses found that AI Overviews place far less emphasis on searcher proximity than local packs do. A business slightly closer to the searcher is marginally more likely to appear, but proximity barely affects ranking order once a business is included. ProRankTracker's testing found the same thing: AI Overviews sometimes cite businesses that aren't even in the 3-pack or on page one.
So the old local SEO playbook (profile, citations, map pack defense) is still necessary, but it no longer decides whether you get named when a customer asks an AI "who should I call.
Why you need to check by city
Two findings make city-level checking non-negotiable.
First, AI Overview content is volatile. Ahrefs found that AI Overview content changes on average every 2.15 days, that consecutive pulls differ 70% of the time, and that 45.5% of citations change when an overview updates. The answer stays semantically the same, but the sources underneath it shuffle constantly. One manual search tells you almost nothing. A pattern across weeks tells you everything.
Second, your visibility genuinely differs by market. Search Engine Land found that AI Overview prominence for a given industry is consistent across cities, but which businesses get named is not. If you serve five cities, you have five different visibility problems.
The baseline is grim, which is actually good news for anyone willing to work. Local Dominator's August 2026 audit of 12,535 local business domains found the average business is mentioned by just 8.24% of the six AI engines tested. The typical local business is invisible to roughly 92% of AI engines checked. Even businesses that do get mentioned average only a 19% share of voice, splitting the answer with four or five competitors.
How to check your visibility by city
Step 1: Build your query list
Write down 20 to 30 queries a customer might actually ask an AI about your service and city. Mix the types:
- Commercial: "best HVAC company in [city]", "emergency plumber near me"
- Problem-based: "why is my AC leaking water"
- Pricing: "cost of roof replacement in [city]"
- Comparison: "[your service] vs [alternative] in [city]"
- Attribute-specific: "dentist open Saturdays in [city]", "restaurant with patio seating [city]"
Longer, conversational queries (7+ words) trigger AI Overviews 46% of the time, so include a few phrased the way people actually talk.
Step 2: Run the queries per city, and log four outcomes
For each query, in each city, record which of four states you're in. ProRankTracker's framework is the clearest one I've seen:
- Competitors named, you're absent (worst case)
- Your site cited as a source link, but your brand never named in text
- Your brand named in text but not linked
- Named in text and cited as a linked source (best case)
The distinction matters. An AI can link your page without recommending you. Being named in the answer's prose is what gets you the call.
Step 3: Calculate share of model
Share of Model = queries where you're mentioned ÷ total queries. Track it monthly, per city. Most local businesses start below 10%. When you do appear, note which sources the AI cited: your website, your Google Business Profile, a Yelp review, a local news article, a "best of" listicle. That tells you which channels are actually feeding your visibility so you can double down.
Step 4: Repeat monthly, not once
Because citations churn every couple of days, run lightweight weekly spot-checks on your 5-10 highest-priority queries and a full monthly audit of the whole list. A single screenshot of an AI answer is a rumor. A monthly spreadsheet is evidence.
Tools that track local AI visibility for you
Manual audits work, but they don't scale past two or three cities. These tools automate the query-running part.
| Tool | What it does for local AI visibility | Starting price |
|---|---|---|
| Local Falcon | Geo-grid rank tracking plus AI visibility scans for AI Overviews, AI Mode, ChatGPT, Gemini, and more, from the same credit system | $24.99/mo (credit-based) |
| Local Dominator | Geo-grid tracking with AI visibility (ChatGPT, Gemini, AI Mode, Perplexity, Claude, Grok) built into the same dashboard | $39/mo |
| Promptwatch | AI visibility tracking with country, state, and city-level tracking, citation source analytics, and crawler logs showing whether Google's AI bots can actually read your pages | $95/mo |
| BrightLocal | Local SEO suite with AI visibility reporting for multi-location businesses | ~$39/mo |
| Whitespark | Local rank tracking and citation tools with AI Overview monitoring | ~$35/mo |
| ProRankTracker | Budget rank tracker with AI tracking (AI Overviews, AI Mode, ChatGPT, Gemini, Copilot, Perplexity) on all paid plans | ~$13/mo |
A few notes from having looked at these closely. Local Falcon's credit system means a 5x5 geo-grid scan costs 25 credits, so price out your actual scan volume before committing. Promptwatch is the strongest option if you want to know why you're invisible, not just that you are: its crawler logs show when Google's AI agents hit your pages and whether they hit errors, and its city-level tracking covers AI Overviews, AI Mode, ChatGPT, Perplexity, and the other engines from one dashboard.
If you're evaluating a wider set of options, the AI rank tracking tools directory at ai-rank-tools.com has a fuller catalog with reviews and comparisons.
How to improve your visibility, city by city
Get complete inside Google's own ecosystem
This is the highest-leverage move, and it's backed by data most people haven't seen. Promptwatch's June 2026 analysis of Google AI Overview citations found that google.com's own share of citations sextupled in five months, from 0.37% in January to 2.30% in June. Add YouTube at 4.16% and Reddit at 2.72%, and Google's own ecosystem plus video accounts for over 9% of all AI Overview citations, roughly double their January share.
The practical takeaway: when Google cites its own properties, you can't win that citation with your website. You win it by being well-represented inside those properties. For a local business, that means:
- A fully completed Google Business Profile, every attribute filled in (each attribute is a filter the AI can match against queries like "emergency service" or "wheelchair accessible")
- Photos that prove geographic legitimacy, landmarks and all
- A YouTube channel with even modest local content, since Google leans on video far more than ChatGPT does
Digital Applied's 2026 compilation found 67% of AI Overviews for local queries reference Google Business Profile data directly, and businesses that get AI recommendations see 2.4x more profile views. Your GBP is no longer just a map pin. It's a citation source.
Fix entity consistency before anything else
Google's AI selects local businesses based on three signals, per ProRankTracker's breakdown: entity clarity (an unambiguous understanding of who you are, what you do, where you are), corroboration (independent third parties confirming your claims), and consistency (matching details across your site, GBP, and every directory). Conflicting NAP details across Google, Yelp, Apple Maps, and your industry directories don't just cost you rankings. They lower the AI's confidence in your entity, which can disqualify you from the answer entirely.
The minimum platform list: Google Business Profile, Yelp, Facebook, Bing Places, Apple Maps, Yellow Pages, BBB, your top 2-3 industry directories, your local chamber of commerce, and the data aggregators. A chamber of commerce listing carries more weight than 20 generic directories combined.
Build reviews that mention specifics
AI reads review text as evidence. A review saying "they came out same-day and fixed the AC in two hours" gives the AI verifiable attributes to match against "emergency AC repair" queries. A review saying "great company, 5 stars" gives it nothing. Ask happy customers to mention the service, the speed, the neighborhood. And respond to every review, positive or negative. A 4.6 rating with a few well-handled negative reviews reads as more trustworthy to filtering systems than a suspiciously perfect 5.0.
Write location pages with fact density, not keyword padding
Promptwatch's July 2026 data on what AI Overviews cite shows a shift worth acting on: product pages overtook listicles as the most-cited content format in late July, the first time that's happened. For a local business, your service and location pages function like product pages. AI search is warming to brand-owned commercial pages, which means your own pages have a real shot at the citation.
What gets cited is structured, specific, directly useful information:
- NAP, service area, and hours in table format
- City-specific testimonials with neighborhood references
- Local regulations, pricing ranges, and landmark photos with descriptive alt text
- FAQs answering real pre-sale questions
Templated city pages with swapped keywords get filtered out. A page that reads as local because it actually is local gets cited.
Schema: do it, but don't expect magic
Here's a myth worth killing. The common claim is that adding schema markup boosts AI citations. Ahrefs ran a controlled study tracking 1,885 pages that added schema and found citations barely moved in aggregate. Schema helps Google understand your entity, and LocalBusiness markup appears on fewer than 4% of mobile pages, so implementing it correctly is still a differentiation opportunity. But it's plumbing, not a strategy. Use the most specific type (Restaurant, Dentist, Plumber) rather than generic LocalBusiness, include areaServed if you're a service-area business, and make sure the markup matches your visible text. Then put your actual effort into content and reviews.
Common mistakes that keep local businesses out of AI answers
- Checking once and concluding. With citations churning every two days, one search is noise.
- Treating map pack rank as AI visibility. They're different signal sets. A #2 map pack position tells you nothing about whether the AI names you.
- Sparse Google Business Profile. Missing attributes are missing query matches.
- Inconsistent NAP across directories, which erodes entity confidence.
- Vague reviews that name no services or attributes.
- Thin templated city pages with no local specifics.
- Relying on schema alone and calling it GEO.
- Ignoring Google's self-citations. If Google cites its own properties for your queries, you need presence inside those properties, not just a better website.
Measuring whether it's working
Beyond your monthly share-of-model spreadsheet, watch these:
- Branded search growth in Google Search Console (people searching your name after the AI mentioned you)
- GA4 referral traffic from AI domains, bucketed into a custom channel group
- Google Business Profile insights: calls, direction requests, profile views
- Review velocity and the specificity of review language
- Which citation sources appear when you do get named, per city
Expect 2-4 months of sustained effort before visibility moves, the same timeline as traditional SEO. The first gains usually come from NAP consistency and profile completeness. The compounding gains come from review velocity, third-party mentions, and location content built over time.
The bottom line
AI Overviews have added a visibility layer above the local pack, built from entity consistency, corroboration, and content clarity rather than proximity. Most local businesses are invisible in it right now, which means the bar for standing out is still low. Build the query list, run it monthly per city, log the outcomes, and fix what the logs tell you is missing. The businesses doing this in 2026 will be the incumbents everyone else is trying to displace in 2028.