Key takeaways
- AI Overviews now appear on roughly 68% of local service queries, but local packs still dominate simple "near me" searches (90%+ appearance rate), so this is a both/and problem, not an either/or one.
- AI platforms are far more selective than Google's map pack: ChatGPT recommends only about 1.2% of business locations for a given category, Perplexity 7.4%, Gemini around 11%, versus 35.9% visibility in Google's local 3-pack.
- Gemini's recommendations are roughly 10x more accurate than ChatGPT's because it's grounded directly in Google Maps data (100% profile accuracy vs. 68% for ChatGPT and Perplexity).
- Review quality has a hard cutoff, not a gradient: businesses near 3.4 stars with sub-5% response rates get excluded from AI recommendations entirely, not just ranked lower.
- NAP consistency, a complete Google Business Profile, and fresh, specific on-page content matter more than backlinks or raw review volume for getting named by name.
Why this question matters more than it used to
I'll be honest, a year ago I would have told a plumber or a dentist that "AI search" was a nice-to-have, something to think about after the website and the Google Business Profile were sorted. That advice doesn't hold anymore. When someone asks "best HVAC company near me" today, a growing share of the time they get a written answer naming one to three businesses before they ever scroll to a map or a list of blue links.
The scale of the shift depends a lot on which platform you're looking at. Local AEO case study data from Whitespark, based on 540 queries across Houston, Phoenix, and Denver in six industries, found Google AI Overviews appearing on an average of 68% of local-type queries, with a range of 57% to 80% depending on the industry. Local packs, by comparison, showed up in only 39% of those same queries. That's a real reversal of what used to be the default local search experience.
But here's the nuance that gets lost in a lot of the hype: AI Overview prevalence depends heavily on query phrasing, not just topic. The same Whitespark data shows AI Overviews appearing for only 15% of simple local-intent queries like "tacos san francisco," but 92% of informational queries like "how long does an eye exam take near me" and 97% of hybrid queries like "average cost of dental implants in phoenix." Simple, short local queries still trigger the traditional map pack most of the time. It's the longer, more specific, more conversational questions where AI answers take over.
AI assistants are far more selective than Google's map pack
This is the part that should change how local businesses think about their odds. SOCi's 2026 Local Visibility Index, which looked at roughly 350,000 business locations, found that AI platforms recommend a much smaller slice of businesses than Google's local 3-pack does. ChatGPT recommended only about 1.2% of locations in its dataset. Perplexity recommended 7.4%. Gemini recommended about 11%. Compare that to the 35.9% visibility rate businesses get in Google's standard local 3-pack, and you start to see why SOCi describes AI search as roughly 30 times more selective than traditional Google search.
Why such a gap between platforms? Data accuracy is the likely driver. The same SOCi research found business profile accuracy at only 68% on both ChatGPT and Perplexity, but 100% on Gemini, because Gemini is grounded directly in Google Maps data rather than pulling from a patchwork of indexed web pages. That's roughly why Gemini's recommendation rate runs about 10 times higher than ChatGPT's. If your Google Business Profile is clean, Gemini already has a head start on recommending you. If it isn't, you're fighting an uphill battle on every platform that touches that data.
What decides whether an AI names your business
Entity consistency, not backlinks
Traditional local SEO rewarded businesses that accumulated backlinks and citations over years. AI systems care less about link volume and more about whether they can confirm, with confidence, that your business is a single consistent entity. If your name shows up as "Smith Plumbing," "Smith Plumbing LLC," and "Smith Plumbing & Heating" across your website, Google Business Profile, Yelp, and Bing Places, that's three slightly different entities in the eyes of a model trying to verify who you are. Multiple sources cite NAP (name, address, phone) inconsistency as the single most common and most damaging technical mistake local businesses make, and it's almost always an accident rather than a strategy choice.
Reviews have a cutoff, not a slope
This one surprised me a little. The SOCi data found that locations recommended by ChatGPT average 4.3 stars, while businesses near 3.4 stars with review response rates below 5% are effectively excluded from recommendations altogether, not just pushed lower in a ranked list. That's a meaningfully different mental model than classic SEO, where a lower rating usually just meant a lower position. Here it can mean you don't exist in the AI's answer at all. Responding to reviews, including the negative ones, isn't just a reputation exercise anymore. It's a visibility requirement.
Your own website carries more weight than you'd think
One widely cited interpretation of the SOCi dataset puts the figure at around 44% of ChatGPT's local-business information coming from the business's own website, with only about 2% coming from forums like Reddit. I'd treat that specific percentage as a secondary interpretation rather than gospel, but the direction matches what we're seeing more broadly in AI citation data. Promptwatch's research on ChatGPT's citation share by domain rank found mid-authority domains (DR 46-75) earning almost half of all ChatGPT citations in August 2026, while the very top authority tier actually fell behind lower-authority sites by month's end. The practical read for a local business: you don't need Domain Rating 90 to get cited. You need pages that answer the question clearly, with specifics, in your own words.
Content type matters, and it's not what classic SEO trained you to expect
According to Promptwatch's data on ChatGPT's citation share by content type, for commercial and comparison-style prompts specifically, which is the closest proxy available to "best [service] near me" behavior, ChatGPT leans heavily on product pages (26.0%), landing pages (20.6%), listicles (15.0%), and direct comparisons (7.9%), adding up to nearly 70% of citations combined. For local service businesses this translates fairly directly: dedicated pages for each service and each city you serve, written in plain language with concrete details, outperform a single generic "Services" page that tries to cover everything at once.
Interestingly, reviews only account for 1.7% of citations in the general dataset, but that number jumps to 5.6%, more than triple, once a prompt names a specific business. So reviews matter most once someone is already evaluating you by name, which is exactly the moment branded search and reputation management intersect with AI visibility.
Freshness is a real ranking lever now
SOCi's analysis of over 129,000 ChatGPT citations found that content updated within the last 30 days received 3.2 times more citations than older material. That's a big enough gap that "set it and forget it" local pages are a genuine liability. Seasonal hours, updated service areas, new pricing, recent before-and-after photos, these aren't just nice touches, they're signals that keep a page in the citation rotation.
Platform-by-platform differences local businesses should actually plan around
| Platform | Grounding source | Recommendation rate | What matters most |
|---|---|---|---|
| Gemini | Google Maps data directly | ~11% of locations, ~10x ChatGPT's rate | Clean, complete Google Business Profile |
| ChatGPT | Web crawl + search, limited local data | ~1.2% of locations | Own website content, entity consistency, reviews once named |
| Perplexity | Web search with heavier social weighting | ~7.4% of locations | Balanced mix of website and third-party mentions |
| Google AI Overviews | Google index + Maps + social | Appears on ~68% of local queries on average | Everything above, since it blends sources |
One more platform split worth knowing: social media citation share varies enormously by engine. Promptwatch's social media citations by AI model data from October 2026 shows ChatGPT citing social platforms in just 0.46% of responses combined, down sharply as Reddit specifically collapsed from a 5.19% share to 0.09%. Google AI Overviews is the opposite story: social platforms make up 12.2% of its citations, about 26 times ChatGPT's share, with YouTube alone at 4.28% and Facebook and Reddit each around 2%. If you're optimizing for AI Overviews, a YouTube presence and active Facebook page are a real citation surface. If you're optimizing for ChatGPT, that same effort won't move the needle much. Same goes for Perplexity, which sits in between at about 2.7% social share.
And one caveat worth keeping in mind before any of this feels like an emergency: Sparktoro's August 2025 research, cited by Search Engine Land, found that only about 20% of Americans use AI tools ten or more times a month, while 95% still use traditional search engines monthly. Semrush has also found that when people start using ChatGPT, their Google usage tends to go up, not down. AI search is additive right now, not a replacement. Treat it as a new channel to win, not as the only channel that matters.
A practical checklist for local businesses
- Audit NAP consistency across your Google Business Profile, Apple Maps, Yelp, Bing Places, and any category-specific directories. Start with the highest-authority platforms and work down.
- Check that your LocalBusiness schema is implemented and accurate. Google's Rich Results Test will tell you immediately if it's missing or broken.
- Build one dedicated page per service and per city or neighborhood you serve, rather than a single page that tries to cover everything. Answer the actual question in the first 40-70 words with a concrete number where possible.
- Respond to every review, especially the negative ones. Given the hard cutoff around 3.4 stars and low response rates, this is no longer optional reputation management, it's a visibility gate.
- Refresh hours, service areas, pricing, and photos on a real schedule. Freshness measurably correlates with citation rate.
- If YouTube or Facebook content makes sense for your business, invest there for AI Overviews specifically. Don't bother chasing Reddit visibility for ChatGPT, that door has mostly closed.
- Query the assistants yourself, monthly, with the exact phrases your customers use. It's a blunt instrument, but it's the most direct way to see whether you're actually being named.
Where AI visibility tools fit into this
Manually asking ChatGPT and Gemini the same ten questions every month works, but it doesn't scale once you're tracking multiple services, multiple locations, or watching what competitors are doing. This is the gap that AI visibility platforms are built to close, tracking how often and in what context a business gets named across ChatGPT, Gemini, Perplexity, and Google's AI surfaces, and flagging what to fix when it isn't happening.
Promptwatch is built around that full loop rather than just the monitoring piece. It tracks citation sources (including crawler logs showing which pages AI systems actually read and whether they hit errors), measures real AI-driven traffic rather than just mention counts, and includes Content Agents that can draft and publish the kind of service and city pages that tend to get cited, directly into a CMS. For a local business or an agency managing several locations, that combination of "are we visible" and "here's what to fix" matters more than a simple tracker that only answers the first half.
For businesses that want to compare options, the GEO software directory at bestgeosoftware.com is a reasonable place to see how different visibility platforms stack up on features and price before committing to one.
The honest bottom line
Local AI search in 2026 rewards operational discipline more than creativity. Fixing three phone numbers across directories, filling out a Google Business Profile completely, writing one page per service per city instead of a single catch-all page, these aren't exciting tasks, but they're the ones that actually move the needle according to every piece of data above. The businesses that get named by ChatGPT or Gemini right now generally aren't the flashiest, they're the ones that made themselves easy to verify. That's a solvable problem, and it's one most competitors in any given local market still haven't solved, which is exactly why it's worth solving now.