Ask ChatGPT for the best local business in a city and its answer is often built from Google Business Profile data. Here is exactly what on your GBP shapes AI recommendations, and the priority list to fix it.
Ask ChatGPT "best movers in [your city]" or "wedding photographer near me" and watch what comes back. When it runs a web search to answer, the recommendations it puts together are often built from Google Maps and Google Business Profile listings: the business names, the categories, the review snippets, even the words customers used in reviews. We see this pattern all the time in our citation audits. Your Google Business Profile is no longer just your Google storefront. It is your AI storefront too.
That changes what "a well-optimized GBP" means. The general mechanics of how ChatGPT picks businesses are covered in our guide to how ChatGPT recommends businesses. This post covers a narrower question: which parts of your GBP actually shape what AI engines say about you, and in what order you should fix them. Categories, attributes, review text, photos, Q&A, services, and hours all carry different weight in an AI answer, and most owners are polishing the wrong ones.
If you want to see what AI engines say right now when someone asks for your service in your area, get a free AI visibility audit and we will run the prompts for you.
TL;DR
- In our October 2026 test of 120 AI answers across eight cities, Google Maps top-20 businesses with 500 or more Google reviews were named by at least one AI 54.8% of the time, against 12.2% for those with under 50 reviews, while average star ratings barely differed (4.82 for named businesses, 4.80 for the rest).
- Your GBP feeds AI answers directly in Google AI Mode, which cited Google pages in 58.3% of its answers in our test, and indirectly in ChatGPT, Perplexity, Gemini and Claude, which cited no Google pages but often repeated ratings and review counts from review sites and directories.
- Your primary category is the strongest filter. AI engines shortlist businesses by category first, and a wrong or vague category leaves you out entirely.
- AI answers quote review text, not just star ratings. Reviews that name the specific service and location become the evidence an AI cites when it recommends you.
- Service lists, attributes, and Q&A give AI engines structured facts to repeat: what you do, what you offer, and answers to common questions in your own words.
- Accurate hours matter more than most owners think, because AI engines repeat them with confidence, and confidently wrong hours cost you jobs and reviews.
- Priority order: category, reviews with service keywords, complete service list, attributes, Q&A, photos, hours. Work top down.
Why your GBP is now your AI storefront
Here is the mechanic in brief. When an AI engine answers a local question, it needs a trustworthy, structured source of local business data. The most complete one on the web is Google's business index: verified listings with categories, locations, reviews, hours, and services, all in a consistent format. So when ChatGPT's web search surfaces local results, Google Maps listings and the pages built from them feature heavily in what it reads. Perplexity and Copilot lean on the same kind of structured local data through their own retrieval. Gemini and Google AI Overviews read Google's business data natively.
The result is blunt. The profile fields you filled in years ago for map rankings are now raw material for AI answers. But AI engines use those fields differently from the Maps algorithm. Maps ranks you by relevance, distance, and prominence. An AI engine writes a sentence about you, which means it needs quotable facts and evidence. That difference is what the rest of this post unpacks, field by field.
One caveat before the list: none of this replaces your website. AI engines cross-check the profile against your site and other sources, and a strong GBP pointing to a weak site is half a foundation. Treat the profile as one pillar of wider generative engine optimization (GEO) work.
| GBP field | What the AI uses it for | Weight in AI answers |
|---|---|---|
| Primary category | Deciding whether you make the shortlist at all | Highest |
| Review text | Evidence for why it recommends you | Very high |
| Services list | What it says you do | High |
| Attributes | Matching qualifiers in the question ("weekend", "free estimates") | Medium to high |
| Q&A | First-party answers it can repeat | Medium |
| Photos | Proof you are a real, active business | Medium |
| Hours | Facts it states flatly to customers | Medium, high risk if wrong |
What our data shows about Google listings and AI picks
In October 2026 we asked ChatGPT, Perplexity, Gemini, Claude and Google AI Mode to name the best moving company, wedding photographer and solar installer in eight cities, then matched every pick against Google's top 20 map results for the same search (full study here). That gave us 480 Google Maps listings, each with its rating, review count, map position, claimed status and website link, and a clear answer on whether any AI named it. Those five listing fields are all the Google Business Profile data the test captured, so the findings below cover them and nothing else.
Do more Google reviews get you named by AI?
Share of Google Maps top-20 businesses named by at least one AI, by review count
Businesses with 500+ Google reviews were named 4.5 times as often as businesses with under 50 (54.8% vs 12.2%).
- Under 50 reviews12.2%
196 listings
- 50-19924.2%
149 listings
- 200-49935.5%
62 listings
- 500+54.8%
73 listings
Source: Orkkid AI Recommendation Study, October 2026 (480 Google Maps listings)
Review count is the listing field that tracks AI recommendations most closely. The share of businesses named by at least one AI rises at every step, from 12.2% under 50 reviews to 54.8% at 500 or more. If you are sitting on a few dozen reviews, asking every finished job for one is the profile work our data backs most clearly.
Reviews and rating of every Google Maps top-20 business
Hover a dot to see the business. Purple dots were named by at least one AI.
Named businesses had a median of 204 Google reviews, against 55 for the businesses AI left out. Star ratings were almost identical.
Source: Orkkid AI Recommendation Study, October 2026 (456 listings with reviews)
Star rating tells a very different story. Named and unnamed businesses averaged 4.82 and 4.80 stars, close enough to be the same. Perfect 5.0 ratings did worst of all: only 19% of the 231 listings at 5.0 were named, and their median was just 51 reviews. A handful of glowing reviews looks good on Maps but gives an AI little evidence to work with, while listings rated 4.8 or 4.9, with medians of 156 and 237 reviews, were named 45% of the time.
| Google star rating | Listings | Named by at least one AI | Median Google reviews |
|---|---|---|---|
| Under 4.6 | 54 | 24.1% | 55.5 |
| 4.6 to 4.7 | 60 | 25% | 196.5 |
| 4.8 | 40 | 45% | 156 |
| 4.9 | 71 | 45.1% | 237 |
| 5.0 | 231 | 19% | 51 |
The other 24 listings had no rating at all, and none of them was named by any AI.
Here is how named and unnamed businesses compare on every listing field we captured:
| Listing field | Named by at least one AI (122) | Not named (358) |
|---|---|---|
| Median Google reviews | 204 | 55 |
| Average star rating | 4.82 | 4.80 |
| Average Google Maps rank | 9.4 | 10.9 |
| Profile claimed | 95.1% | 95% |
| Listing links to a website | 98.4% | 91.6% |
Claiming your profile is table stakes, since 95% of businesses in both groups had done it. A website link matters more than it looks: only 2 of the 32 listings without one were named, a small sample but a consistent pattern.
The higher a business ranks in Google Maps, the more likely AI names it
Share of Google Maps top-20 businesses named by at least one AI, by map position
37.5% of businesses in the top 3 map spots were named by at least one AI, against 22.5% of those ranked 11 to 20.
- Map rank 1-337.5%
72 listings
- Map rank 4-1024.4%
168 listings
- Map rank 11-2022.5%
240 listings
Source: Orkkid AI Recommendation Study, October 2026 (480 Google Maps listings)
Map rank helps, but less than you might expect. Businesses in the top three map spots were named 37.5% of the time, against 22.5% for those ranked 11 to 20, and most of the AI picks we recorded came from outside the map top 20 entirely. Ranking well on Maps is worth having, but it does not secure the AI recommendation on its own.
Where the engines got their Google data also differed. Google AI Mode cited Google pages in 58.3% of its 24 answers. ChatGPT, Perplexity, Gemini and Claude cited no Google pages at all, yet Google ratings still reached their answers through other sites. ChatGPT, answering for Austin movers, wrote that "AustinTopRated names it among top-rated Austin movers based on Google rating/review volume." Claude, describing Hornet Moving for Charlotte, wrote: "It has a 4.9/5 Google rating across 5,491 reviews and uses per-minute pricing." Your review count travels well beyond Google, because review sites, best-of lists and directories copy it into the pages the AI engines read.
→ Free free GEO audit: see whether ChatGPT and Google AI recommend your business, and who they name instead. Check my AI visibility.
Primary category: the filter that decides if you exist
Before an AI engine weighs anything else, it has to build a shortlist, and the shortlist is built by category. If someone asks for a wedding photographer and your primary category is just "Photographer", you are not losing the comparison. You were never in it.
What to do:
- Make your primary category the most specific one that matches your core service. "Wedding photographer" beats "Photographer" for wedding questions. "Wedding venue" beats "Event venue" for couples looking for a ceremony site. Check which categories the businesses already showing up in AI answers for your target question use, and match their specificity.
- Use secondary categories for every real service line, not aspirational ones. Each legitimate secondary category puts you on another shortlist.
- Never keyword-stuff the business name. AI engines repeat business names word for word, and "Smith Moving | Best Cheap Movers Near You 24/7" reads as spam to a model trained on natural language. It also breaks Google's guidelines.
This is the highest-leverage five minutes on this entire list. We have seen businesses appear in AI answers for a service line within weeks simply because a category change put them in the candidate pool for the first time.
Review text: the evidence AI engines quote
Star ratings get you considered. Review text gets you recommended. This is the part most owners miss about Google Business Profile AI signals: when ChatGPT justifies a recommendation, it often paraphrases or nearly quotes what reviewers wrote. "Customers mention careful handling of furniture and on-time arrival" is a sentence built from review text, not from a star count.
That means the words in your reviews are now marketing copy you do not write. Influence them the legitimate way: through the ask.
- Ask customers to mention the specific job and location. Not scripted, just prompted: "If you have a minute, it helps us if you mention what we did and where." A review saying "moved our three-bedroom house across the state, nothing broken, crew arrived on time" hands the AI a service, a scope, and a proof point in one sentence.
- Favor a steady flow over bursts. AI engines, like Google, treat a steady stream of recent reviews as a sign of a business that is good right now. Twenty reviews from 2023 and silence since reads as decline.
- Reply to every review, especially the bad ones. Replies are part of the retrievable text on your listing. A calm, factual reply to a bad review is often the difference between an AI mentioning a complaint and an AI noting that issues get resolved.
- Never buy or fake reviews. Pattern detection catches it, and the penalty is your whole trust profile.
Our free Google review link generator makes the ask a one-tap link you can text or email after every job.
Services list: the structured menu AI reads back
The services section of your GBP lets you list each service with a name and description. Most businesses leave it half-empty or lump everything into three vague entries. That is a direct loss, because AI engines treat the services list as a structured menu of what you do, and repeat entries from it when a question matches.
Fill it properly:
- One entry per real service, named the way customers ask for it: "long-distance moving", "packing services", "engagement sessions", "solar battery installation". Not internal jargon.
- Write the descriptions. Two or three plain sentences saying what the service includes and who it suits. These descriptions are retrievable text.
- Cover the long tail. Niche services are where AI answers are least contested. A mover that lists "piano moving" as its own service, with a description, gives the AI a reason to name it for that exact question.
Be the business AI recommends
Attributes: small toggles, outsized answers
Attributes are the checkboxes on your profile: wheelchair accessible, free estimates, online estimates, women-owned, veteran-owned, on-site parking. They look trivial. They are not, because they map directly onto the qualifiers people put in AI prompts. "Movers near me that give free in-home estimates" or "wedding venue with on-site parking" are exactly the questions where an attribute includes you or leaves you out.
Check every attribute Google offers for your category and set each one that is true. Then look at the qualifiers your customers actually use, and make sure the matching attributes, services, and review evidence all agree. An attribute claiming weekend availability with no review ever mentioning a weekend job is a weak signal. The two together are a strong one.
Q&A: answer the questions before the AI improvises
The questions and answers section on your profile is public, retrievable text, and it is one of the few places on your GBP where you can publish direct answers in your own words. Left empty, it is a void. Worse, anyone can post a question, and unanswered or community-answered questions become the record.
Seed it yourself, which is allowed: post the eight to ten questions you get on the phone every week, and answer each in two to four sentences. Do you charge a travel fee? Do you serve my area? How far ahead do I need to book? Are you licensed and insured? These are exactly the follow-up questions people ask AI engines, and a first-party answer on a verified profile is a source an engine can lift directly.
Google has been changing how Q&A appears on some profiles. If yours no longer shows the section, put the same questions and answers on your website as an FAQ page with FAQ schema. Our FAQ schema generator builds the markup. Either way, keep answers factual and standalone, the same discipline we teach for content AI will cite.
Photos: proof of a real, operating business
Photos do not get quoted in a text answer, so owners assume they do not matter for AI. They still do, for two reasons. First, multimodal engines increasingly read images, and photos of real jobs, real trucks, and real premises back up that your listing describes an actual operating business. That feeds the trust assessment. Second, photo volume and recency go hand in hand with profile completeness and activity, and complete, active profiles are what engines prefer to recommend.
Keep it simple: upload genuine job photos monthly, name the files sensibly before upload, and make sure your premises, team, and work are all represented. Skip stock photography entirely. A model that has seen the same stock image on four hundred websites learns nothing from it, and a human who clicks through learns worse.
Get your free GEO audit
Buyers now ask AI for a shortlist before they visit a single website. Find out whether your business is on it, and how to get there first.
Check my AI visibilityOpening hours: the fact AI repeats most confidently
AI engines state opening hours as flat fact. "They are open until 5pm today." If your hours are wrong, the engine is confidently sending customers to a closed door, and the resulting one-star "called and nobody answered" review becomes retrievable evidence against you.
Keep base hours exact, set holiday hours ahead of every public holiday (including regional ones your customers may forget), and if you offer after-hours or weekend service, show it in the attributes and services list rather than falsely listing 24-hour opening hours for an office that is not staffed overnight. Accuracy here is trust, and trust is the currency of every AI recommendation a local business earns.
The priority action list
Do these in order. The first three carry most of the weight.
- Fix your primary category to the most specific true match, and add every legitimate secondary category. (Thirty minutes.)
- Build the review engine. Ask every completed job for a review within 24 hours, prompting for service and location mentions. Reply to everything. (Ongoing, and the biggest long-term lever.)
- Complete the services list. Every real service, named in customer language, with a written description. (One to two hours.)
- Set every true attribute, matched to how customers qualify their prompts. (Twenty minutes.)
- Seed the Q&A with your ten most common phone questions and factual answers, or publish them as an FAQ page if Q&A is not available. (One hour.)
- Upload genuine photos monthly and delete anything stock. (Ongoing.)
- Check hours, set holiday hours each quarter, and align after-hours availability across attributes and services. (Twenty minutes, quarterly.)
- Cross-check consistency between your GBP, your website, and your directory listings. Engines verify businesses across sources, and contradictions cost you confidence. Your name, address, phone, and service claims should match everywhere.
Then measure. Once a month, ask ChatGPT, Gemini, and Perplexity the handful of questions that matter to you ("best [your service] in [your city]", plus qualifier variations) and record whether you appear and what evidence the engine cites. Our prompt panel generator builds the list, and our guide to tracking ChatGPT citations covers the method. That monthly panel, alongside your normal GBP insights, tells you whether the work is landing.
What this looks like by industry
The fields that move AI answers vary by category.
For moving companies, review text about care, timing, and honest pricing dominates, because the fear in every moving question is damage and surprise fees. Services entries for long-distance, packing, and specialty items like pianos put you on more shortlists.
For wedding vendors, photos and review depth carry the proof, because couples ask about style and experience. Reviews that name the venue and the type of wedding give the AI exactly the evidence it needs.
For solar companies, service entries for panels, batteries, and commercial work matter most, along with reviews that mention the system installed and the result, because buyers ask comparative, product-specific questions.
Same checklist, different emphasis.
How we got this data
On 7 October 2026 we asked ChatGPT (gpt-5.5), Perplexity (sonar-pro), Gemini (gemini-3.8-flash) and Claude (claude-sonnet-5-5), all with web search on, plus Google AI Mode: "Who is the best {moving company, wedding photographer or solar installer} in {city}? Give me your top picks." We ran it in Denver, Austin, Phoenix, Charlotte, Seattle, Tampa, London and Toronto, which produced 120 answers, 824 recommendations and 558 unique businesses. We pulled the Google Maps top 20 for each industry and city (480 listings, with rating, review count, rank, claimed status and website) and matched every AI pick to it by name, and we recorded and classified every source the engines cited. Each prompt ran once per engine, so treat this as a snapshot: AI answers vary between runs, the map top 20 reflects a city-center search, and the patterns show association, which is weaker evidence than cause. The full method and data are in our AI recommendation study.
Find out if AI recommends your business
Your free audit tests the questions your customers ask across ChatGPT, Perplexity and Google AI, shows which businesses get named today, and lays out what it will take to add yours.
Tested across ChatGPT · Perplexity · Google AI Overviews · Gemini
Check my AI visibilityFrequently asked questions
Does ChatGPT really use Google Business Profile data?
When ChatGPT answers local questions with web search, the results it reads often include Google Maps listings and pages built from Google Business Profile data, because that is the most complete structured source of local business information on the web. We see GBP-derived details, including review phrasing, appear in its answers regularly during citation audits. In our October 2026 test, none of ChatGPT's 24 answers cited a Google page directly, so that data often reaches it second hand, through review sites and directories that repeat your Google rating and review count. Gemini and Google AI Overviews read Google's business data natively.
Which Google Business Profile field matters most for AI recommendations?
Your primary category, because AI engines shortlist businesses by category before weighing anything else. A vague or wrong category leaves you out of answers no matter how strong your reviews are. After category, review text is the biggest lever, since AI answers quote and paraphrase what customers wrote when they justify a recommendation.
Do star ratings or review text matter more to AI engines?
Both matter, but they do different jobs. Star ratings and review volume get you into the trusted pool. Review text supplies the specific evidence an AI uses to describe and recommend you. A 4.8-star business whose reviews name specific services and places will be described more accurately, and recommended more often, than a 4.9-star business with one-line "great service" reviews. In our October 2026 test of 480 Google Maps listings, businesses the AIs named averaged 4.82 stars against 4.80 for the rest, while their median review count was 204 against 55.
Can I optimize my GBP for AI without changing my website?
You can improve your AI visibility meaningfully through the profile alone, especially the category, services, and review work. But engines cross-check your profile against your website and directory listings, so a profile that contradicts a thin or inconsistent site hits a ceiling. To an AI, the profile and the website are one business, and both need to agree. Our guide to ranking in Google AI Overviews covers the website side.
How often should I update my Google Business Profile for AI visibility?
Reviews and replies should be continuous, photos monthly, and a full field check quarterly, including holiday hours before each public holiday. AI engines reward current, active profiles, and retrieval happens in real time, so stale data gets repeated to customers the moment it goes wrong. A 30-minute monthly routine covers it.
Do Google Posts help with AI recommendations?
Posts are a minor signal compared with categories, reviews, and services, but they add activity and fresh retrievable text to your listing. If you have news, offers, or finished projects worth posting, post them. Do not churn out filler posts at the expense of the review and services work higher up the list.
Your Google Business Profile is being read to your future customers, out loud, by machines you do not control. The only question is whether it says what you want it to say. Work through the priority list above, or if you would rather see which prompts your business shows up in today and which fields are holding you back, book a free AI visibility audit and we will map it for you as part of a full GEO review.
