BRYGHTON MEDIA Free Consultation

From the Blog — AI Visibility

How Do You Get ChatGPT to Recommend Your Local Business?

45% of consumers now use ChatGPT or another AI tool for local business recommendations, up from 6% a year earlier (BrightLocal 2026). What an AI assistant looks for before it names a business, what we measured on our own site, and eight steps any small business can run.

Can ChatGPT recommend a local business?

Yes, and people are asking it to. In BrightLocal’s 2026 Local Consumer Review Survey, 45% of consumers said they use ChatGPT or another generative AI tool for local business recommendations, up from 6% a year earlier, which makes AI the third most popular source of recommendations. An assistant can only recommend what it can read: your website, your reviews and the pages that mention you.

This changes the job of a small business website. For twenty years the goal was a blue link on page one. Now a meaningful share of the people who would have typed “flight school near me” or “roofer in Newburgh” into Google are typing a full sentence into an assistant and reading one answer with three or four businesses in it. Being in that answer is what we mean by AI visibility, and it is the reason this blog exists.

How many people are asking AI for recommendations?

More than most owners assume. OpenAI’s chief executive said in October 2025 that ChatGPT had 800 million weekly users, as reported by Business Insider. BrightLocal’s 2026 survey found 40% of consumers trust AI for local searches versus 32% who do not, 42% trust AI recommendations as much as traditional reviews, and 82% read the AI-written review summaries that now sit on top of many business listings.

Two things in those numbers matter for a business with one location. First, the trust is already there for a large minority; this is not a novelty. Second, the AI summary of your reviews is being read even by people who never open an assistant, because Google and the review platforms now generate one at the top of the listing. What is in your reviews is what the summary says about you.

What does an AI assistant look for before it cites a page?

Content it can lift and attribute. The first peer-reviewed study of this, “GEO: Generative Engine Optimization” from Princeton University and collaborators (KDD 2024), tested changes to web pages across 10,000 queries and found the right changes raised a page’s visibility in AI answers by up to 40%. The changes that worked best were adding statistics, citing sources and adding quotations. Keyword stuffing, the old trick, made pages less visible.

Read that as a writing rule. An assistant assembling an answer about flight schools in the Hudson Valley is looking for a paragraph it can quote: a plain statement, a number, a source, a name. A page of adjectives (“premier,” “world-class,” “passionate”) gives it nothing to hold on to.

We tested this on ourselves. In September 2026 we ran an open-source citability scorer over the Bryghton Media homepage. It scored an average of 40.9 out of 100 across 14 content blocks, and the two weakest factors were “statistical density” and “uniqueness signals,” both at zero on 12 of the 14 blocks. The page was well designed and said nothing an assistant could cite. This blog, the two numbered answer blocks now on our homepage, and this post are the fix, and we re-score after each change.

How do you make your business easier for AI to recommend?

Eight steps, in order: let the AI crawlers in, publish an llms.txt file, write answer-first blocks with real numbers and named sources, add structured data, get recent reviews and respond to them, complete your Google Business Profile, put a named person with credentials behind the content, and update the site regularly. None of these replaces traditional SEO; they sit on top of it.

  1. Let the crawlers in. Check robots.txt. The assistants read the web with named crawlers: GPTBot and OAI-SearchBot for OpenAI, ClaudeBot for Anthropic, PerplexityBot, Google-Extended for Gemini. Many sites block them by accident through a plugin default. Ours explicitly allows each one.
  2. Publish an llms.txt. A plain-text file at yoursite.com/llms.txt that says who you are, what you do and where the important pages are. It is an emerging convention, not a ranking factor, and it takes an hour. Ours lists our ten service lines, the case study numbers and every page on the site.
  3. Write answer-first blocks. A question as the heading, a direct 40-to-80-word answer under it, then the detail. Put a real number and its source in the answer. This is the Princeton finding applied one paragraph at a time, and it is how every section of this post is built.
  4. Add structured data. JSON-LD for Organization or LocalBusiness, Person for the owner, Article for posts and FAQPage for question blocks. It tells a machine what the page is instead of making it guess.
  5. Get recent reviews and answer them. BrightLocal: 47% of consumers will not use a business with fewer than 20 reviews, 74% only weigh reviews from the last three months, and 80% are likely to use a business that responds to all of its reviews. The AI summary of your listing is written from these.
  6. Complete the Google Business Profile. Only 35% of small businesses have one at all, per BrightLocal. Hours, photos, services, and a phone number that gets answered.
  7. Put a person behind the content. Stanford’s Web Credibility guidelines, built on research with more than 4,500 people, list “highlight the expertise in your organization” and “show that honest and trustworthy people stand behind your site” among ten things that make a site believable. Assistants weigh the same signals. Every post on this blog carries the author’s name and credentials for that reason.
  8. Update it. Stanford guideline eight is “update your site’s content often.” A blog with a date on it is the simplest way to show a page is alive.

Does AI visibility replace SEO?

No. The assistants still find pages through search indexes, crawlable HTML and links, so a site that is slow, blocked or thin does badly in both. AI visibility is a layer on top of ordinary SEO: the same foundations, plus content written so a machine can quote it and a person can trust the quote. The Princeton paper’s own finding was that keyword stuffing hurt while evidence helped.

The practical difference is in what you write. Traditional SEO copy repeated the target phrase; citable copy states a fact, names a source and moves on. The good news for a small business is that this favors the honest operator with real numbers over the competitor with the bigger content budget.

What did we change on our own site, and what did it score?

Before: a 40.9 average citability score on the homepage, zero statistical density on 12 of 14 blocks, an llms.txt that was an auto-generated file listing dead URLs, and no Article or Person schema. After the September 2026 pass: a rewritten llms.txt, a robots.txt that names every major AI crawler, Organization, Person, Article and FAQPage JSON-LD, two answer blocks on the homepage with sourced numbers, and this blog. Re-scored on 28 September 2026 with the same tool, the preview homepage averages 44.8 out of 100 (from 40.9), and the two new answer blocks score in the 60s and 70s where the old blocks scored in the 20s and 40s; this post averages 56.8 on first scoring. We will keep re-scoring after each change; the point of the exercise is to be measured by the same tool we would use on a client.

Where to start

Ask ChatGPT, Perplexity or Google’s AI Mode the question your customer would ask (“best flight school near Orange County NY,” “who builds websites for contractors in the Hudson Valley”) and read what comes back. If you are not in it, look at who is and open their pages: you will usually find numbers, sources, reviews and a named person. Then run the eight steps above. If you would rather we did it, AI visibility is part of the website and Google Business Profile lines on our What We Offer page, and the free consultation starts with exactly that search.

Sources

  1. BrightLocal, Local Consumer Review Survey 2026 — brightlocal.com/research/local-consumer-review-survey. Figures cited: 97% read reviews; 41% always read them; 47% will not use a business with fewer than 20 reviews; 74% only weigh reviews from the last three months; 54% visit the website after positive reviews (32% in 2019); 89% expect a response; 80% likely to use a business that responds to all reviews; generic replies make 50% unlikely; 83% of people asked to leave a review did; 28% always will if asked; Google share of review reading 71% (83% in 2025); 35% of SMBs have a Google Business Profile; use of ChatGPT and other generative AI tools for local recommendations 45% (6% the year before), third most popular source; 40% trust AI for local searches vs 32% not; 42% trust AI recommendations as much as reviews; 82% read AI review summaries.
  2. Sam Altman, OpenAI DevDay, October 2025, as reported by Business Insider: ChatGPT at 800 million weekly users — businessinsider.com.
  3. Aggarwal et al., “GEO: Generative Engine Optimization,” ACM KDD 2024 (Princeton University and collaborators) — arxiv.org/abs/2311.09735. GEO-bench of 10,000 queries; the best-performing changes raised visibility in generative engine responses by up to 40%.
  4. Stanford Persuasive Technology Lab, Web Credibility Project, Stanford Guidelines for Web Credibility (ten guidelines from three years of research with over 4,500 people) — credibility.stanford.edu/guidelines.
  5. Citability scorer: citability_scorer.py from the open-source geo-seo-claude toolkit (zubair-trabzada/geo-seo-claude, MIT). Our run on bryghtonmedia.com, 28 September 2026: average 40.9/100 across 14 blocks; statistical density and uniqueness signals scored 0 on 12 of 14 blocks.

Questions

AI visibility, answered

What is AI visibility?

AI visibility is whether an AI assistant (ChatGPT, Perplexity, Google’s AI Overviews and AI Mode, Gemini, Copilot) names or cites your business when someone asks it a question you could answer. It depends on what the assistant can read about you: your site, your reviews and pages that mention you.

What is llms.txt?

A plain-text file at the root of a website (yoursite.com/llms.txt) that summarizes who the business is, what it offers and where the key pages are, written for AI systems to read. It is a convention, not a standard with a ranking effect; it costs an hour and removes guesswork.

Should a business block AI crawlers to protect its content?

A business that wants customers usually should not. Blocking GPTBot, ClaudeBot or PerplexityBot removes the site from the answers those tools give. Publishers with paid content have a different calculation; a flight school or a contractor wants to be found.

How long until an AI assistant recommends my business?

Nobody publishes a timeline; each assistant crawls and refreshes on its own schedule. What we do is measure: run a citability scorer on the pages, ask the assistants the customer’s question monthly, and record who is in the answer. The numbers move as reviews, structured data and answer-first content accumulate.

Next Step

Want to know what the assistants say about you?

The free consultation starts with the search your customer would run: in Google, in ChatGPT and in Perplexity. We show you what comes back, who is in the answer instead of you, and what we would change first.

More from the blog

All posts →