The 5 signals that decide whether AI recommends you

“We’re on page one of Google. So why does ChatGPT never mention us?”

It’s a frustrating question. Ranking in search and being recommended by AI are related, but they’re not the same thing. And “being recommended” turns out to have several layers of its own. A business can be named often but described wrongly. It can be described perfectly but only ever appear last on the list. It can be named in ChatGPT and invisible in Google’s AI.

That’s why we don’t measure AI visibility as a single yes or no. We break it into five signals. This article explains each one in plain English: what it means, how it’s measured, and what actually moves it. It’s the thinking behind our AI Visibility Score and the free audit we offer.

If you’re new to the topic, start with What is GEO? A plain-English guide for business owners.

1. What are the five signals of AI visibility?

The five signals are presence (does AI name you), prominence (how strongly it features you), accuracy and sentiment (whether what it says is true and positive), citation strength (which sources it relies on for you), and competitive gap (who it names instead). Together they describe how AI treats your business.

A quick but important point before we go through them. These five signals are measurements of what AI actually does, not secret ranking factors. No AI company publishes the formula it uses to pick businesses. What we can do is measure how AI treats a business, consistently and repeatedly, and then connect each result to the causes that published research points to.

Why five signals and not one? Because the problems behind each one are different, and so are the fixes. A business named in most answers but with the wrong phone number has an accuracy problem that can often be fixed in weeks. A business with perfect details that AI rarely mentions has a presence problem, which takes longer. One number would hide the difference.

Why measure each engine separately? Because they behave differently, even when they come from the same company. One Ahrefs analysis found that Google’s AI Mode and AI Overviews cited the same web pages only 13.7% of the time for the same question. You can be strong in one and missing from another.

Why many questions and many runs? Because AI answers are unstable. A SparkToro and Gumshoe study ran the same prompts through ChatGPT, Claude and Google’s AI nearly 3,000 times. The same list of recommendations came back less than once in 100 runs. A single answer is an anecdote. Many answers make a measurement.

2. Signal 1: Presence. Does AI name you at all?

Presence is the share of relevant customer questions where AI names your business. It’s the foundation, because if AI doesn’t mention you, nothing else matters. Since answers vary between runs, presence is a percentage measured across many questions and repeated runs, never a simple yes or no.

How it’s measured. We build a set of real buying questions for your business, the kind a customer would actually ask, like “who’s the best [service] in [area]” or “who can fix [problem] today.” Each question is run several times in each AI engine. Presence is the percentage of those answers that name you. You can get a rough version yourself with our 10-minute self-check.

What moves it. Two things matter most. The first is a clear identity. AI needs to recognise your business as one consistent thing, with the same name, address, services and contact details everywhere it looks.

The second is being talked about across the web. In an Ahrefs study of 75,000 brands, the number of times a brand was mentioned across the web was the factor most closely linked to appearing in Google’s AI Overviews. It scored 0.664 on their correlation scale, against just 0.218 for backlinks. The study also found that brands in the bottom half for web mentions barely appeared in AI Overviews at all. A later Ahrefs analysis found mentions on YouTube were even more closely linked to AI visibility.

One caution: these are correlations. Businesses that are widely talked about also tend to be bigger and better known, so mentions aren’t a lever you pull in isolation. But the direction is clear. AI favours businesses that the rest of the web already talks about.

For local businesses, those mentions come mostly from reviews, directories and local listings. BrightLocal advises businesses to keep listings accurate and to check the sources AI tools reference, because AI is now a major way consumers find local businesses.

3. Signal 2: Prominence. When AI names you, do you stand out?

Prominence measures how strongly AI features your business when it does mention you. Are you the first recommendation or the last? Does AI give a reason to choose you, or just list your name? Because the order of names shuffles between runs, prominence is an average across many answers.

How it’s measured. For every answer that names you, we look at where you appear, how much the AI says about you, and how it frames you. Being called “a top choice for emergency call-outs” is very different from “you could also consider.” Then we average across all runs.

This idea has a research pedigree. The Princeton study that coined the term GEO measured visibility partly by how many words an AI answer devoted to a source, weighted by how early in the answer it appeared. In plain terms, it’s being mentioned early and at length.

The averaging matters because single positions are unreliable. In the SparkToro study, the chance of getting the same list in the same order was roughly 1 in 1,000. Anyone who tells you “you rank third in ChatGPT” is reporting one roll of the dice.

What moves it. Give AI a specific reason to choose you, and make sure that reason shows up everywhere. “Quality service at fair prices” gives AI nothing to repeat. “Fixed-price quotes before we start” or “specialists in restaurant accounting” does. State it on your website and your Google Business Profile, and ideally hear it echoed back in your reviews.

Reviews matter here because AI looks for consistent themes, just as customers do. In BrightLocal’s 2026 survey, the review factor consumers rated most important was whether a review was backed up by others with a similar sentiment. When ten reviews all praise the same thing, AI has a clear, repeatable reason to feature you.

The research supports writing that’s specific and well supported. In the Princeton study, adding statistics and quotations to content were among the most effective ways to raise its visibility in AI answers. More on this in How to write content AI actually quotes.

4. Signal 3: Accuracy and sentiment. Is what AI says true, and good?

Accuracy checks whether the facts AI gives about your business are correct, including location, services, prices, hours and contact details. Sentiment checks whether the overall picture is positive, neutral or negative. A mention with the wrong phone number, or with a warning attached, can cost you more than no mention at all.

How it’s measured. Every mention is checked against your real details. Wrong addresses, discontinued services, outdated prices and mix-ups with similarly named businesses all get logged. Then we look at tone. Does AI describe you warmly, neutrally, or with caveats like “some customers report delays”?

Why it matters. AI tools get facts wrong more often than most people assume. In a large international study coordinated by the European Broadcasting Union and led by the BBC, journalists reviewed more than 3,000 AI answers to news questions. 45% had at least one significant issue, and 20% contained major accuracy problems, including invented or outdated details. That study was about news, not businesses, but the mechanism is the same. AI repeats what it finds.

An Ahrefs experiment showed how far that goes. A researcher invented a fake company and planted false claims about it online. Nearly every AI tool repeated those claims as fact, even though the company’s own FAQ explicitly denied them. Your website alone can’t outvote the rest of the web.

Sentiment is shaped heavily by reviews, and AI is increasingly how people read them. BrightLocal found that 82% of consumers read AI-generated summaries of reviews, and 23% would base their decision on the summary alone. The AI’s one-paragraph take on your reviews is becoming your first impression.

What moves it. Correct wrong facts at their source, not just on your website. That means old directory listings, outdated articles and forgotten social profiles. Keep your core details identical everywhere. Reply to reviews, personally. In BrightLocal’s survey, 89% of consumers expect businesses to respond to reviews, and generic templated replies put off half of them. And be honest with yourself about recurring complaints. If reviews keep mentioning the same problem, AI will too. Sometimes the fix is in the business, not the marketing.

5. Signal 4: Citation strength. Which sources does AI trust about you?

Citation strength looks at the sources AI draws on when it talks about your business, and whether it cites your own pages. Strong, relevant, accurate sources make your mentions more stable and more persuasive. Being cited from an outdated directory listing is a weakness waiting to happen.

How it’s measured. For every answer that mentions you, we record which websites the AI cites. How often is your own site among them? Are the other sources reputable and relevant to your industry and market? Are they accurate and current? And which sources does AI cite for your competitors that don’t mention you at all?

Why it matters. Being cited appears to pay off. A Seer Interactive study of more than 3,000 search queries found that brands cited in Google’s AI Overviews earned 35% more organic clicks and 91% more paid clicks than brands that weren’t cited. The researchers were careful to say they can’t prove the citation causes the extra clicks. Stronger brands may simply be more likely to get cited. Either way, the brands winning citations were also winning clicks.

It also matters which sources AI trusts. Pew Research Center found that Wikipedia, YouTube and Reddit were the most cited sites in Google’s AI summaries. For local businesses, review platforms and directories often play that role, and which ones matter varies by country.

What moves it. Make your own pages the best source on your topic. Answer common questions directly, with specific, verifiable details, so AI has something worth citing. Then earn a presence on the third-party sources AI already trusts in your market, such as review platforms, industry directories, local press and community forums. AI search in your market: SA, US, UK and Australia covers which sources matter where. Finally, audit the pages AI already cites about you and make sure every one of them is accurate.

6. Signal 5: Competitive gap. Who does AI name instead of you?

Competitive gap measures how your AI visibility compares with the businesses AI recommends instead. It shows who is winning the answers you want, how far ahead they are, and, most usefully, which sources and content give them the edge. It turns a vague worry into a specific to-do list.

How it’s measured. Across the same set of questions, we track every business AI names, not just you. That gives each competitor a share of the answers. Then we dig into why. Which sources does AI cite for them? Which questions do they win that you don’t? What does AI say about them that it doesn’t say about you?

Why it matters. AI recommendations aren’t as random as individual answers suggest. In the SparkToro study, even though the exact lists changed constantly, the tools kept drawing from a fairly consistent group of brands. Think of it as a shortlist that AI keeps reshuffling. Your goal is to get onto that shortlist.

The gaps also tend to compound. In the Ahrefs study of 75,000 brands, those in the top quarter for web mentions averaged 169 mentions in AI Overviews. The next quarter down averaged just 14. Visibility breeds more visibility, which is a good reason to close gaps early rather than late.

One thing can surprise owners. The businesses AI names instead of you aren’t always your real-world rivals. Sometimes it’s a national chain, a directory, or even a business that no longer exists. The SparkToro researchers found AI tools recommending defunct businesses and inactive accounts. Knowing exactly who you’re up against in AI answers can be the first real insight of an audit.

What moves it. Close the three gaps the analysis reveals. The first is the source gap: listings, reviews and articles your competitors appear in and you don’t. The second is the content gap: questions they answer clearly and you don’t. The third is the reputation gap: themes in their reviews that make AI confident recommending them.

7. How do the five signals become one AI Visibility Score?

Each signal is scored separately in each AI engine, using the same set of real customer questions, run several times. Those scores are combined into a single number from 0 to 100. The total tells you where you stand. The breakdown by signal and engine tells you what to fix first.

A single score is useful for two things. It gives you a baseline, and it lets you track progress over time and against competitors in the same market.

But the breakdown is where the value is. Two businesses can have the same total and need completely different work. Take two made-up examples. The first is named often, but AI keeps quoting an old address and a service it no longer offers. Its fix is mostly accuracy work, which can move quickly. The second is described perfectly on the rare occasions it’s named. Its fix is presence work, meaning more mentions, more reviews and more sources, and that takes longer. Same score, very different plans.

The engine breakdown matters just as much. A business can show up well in one AI tool and barely register in another, because each draws on different sources.

A fair caveat: any score is a model, not an absolute truth. It’s most meaningful when you compare it with your competitors in the same market and with your own score over time. That’s exactly how we use it.

What to do this week

• Estimate your presence. Run the 10-minute self-check and count how many answers name you.
• Audit your accuracy. List every wrong fact AI gave about your business, and where each one might come from.
• Write your one reason. In one sentence, why should a customer choose you over the alternatives? Put that sentence on your homepage and your Google Business Profile.
• Reply to your last ten reviews, personally and specifically.
• Map one competitor’s sources. Note which sites AI cites when it recommends them, and claim or complete your listing on one of them.
• Answer one question you’re missing from. Pick a customer question where competitors get named and you don’t, and write a clear, direct answer on your website.

Frequently asked questions

Which of the five signals matters most?

Presence comes first, because AI can’t recommend a business it never mentions. But if AI is already naming you with wrong details, accuracy becomes the most urgent. A wrong phone number or address sends customers away at the very moment they’re ready to call.

Can I measure these signals myself?

Roughly, yes. Our 10-minute self-check gives you a basic read on presence, accuracy and your competitors. Measuring all five properly takes many more questions and repeated runs across every engine, because AI answers change so much from one run to the next.

Why do my results differ between ChatGPT and Google’s AI?

Each engine draws on different sources and weighs them differently. Even Google’s own AI features differ: one Ahrefs analysis found AI Mode and AI Overviews cited the same pages only 13.7% of the time for the same question. That’s why each engine is scored separately.

Do backlinks still matter for AI visibility?

They still matter for traditional SEO, which feeds AI tools that search the web. But in Ahrefs’ study of 75,000 brands, mentions across the web were far more closely linked to AI visibility than backlinks were. Being talked about matters more than being linked to.

Can a small business compete with big brands in AI answers?

Yes, especially in local and specialist markets, where AI looks for the best fit for a specific question. The Princeton GEO study found that lower-ranked websites gained the most from clear, well-supported content. One method raised visibility by 115.1% for sites ranked fifth in search results.

What’s a good AI Visibility Score?

It depends on your market. A score is most useful compared with your direct competitors for the same customer questions, and with your own score over time. Rising steadily while closing the gap on the leaders matters more than hitting any particular number.

How long does it take to improve these signals?

Accuracy fixes, like correcting wrong listings, can show up in AI answers within weeks. Presence and citation strength take longer, usually months, because they depend on reviews, mentions and content building up across the web. Tracking the signals monthly shows you what’s moving.

Find out your five signals

Our free audit scores your business across all five signals, in every major AI engine, using real questions your customers ask. You’ll see where you stand, who AI names instead of you, and which fixes will make the biggest difference.

Book a free discovery call and choose the audit that fits: AI Visibility & GEO, SEO, or Local Business SEO. No pitch until you’ve seen the gap.

Book a discovery call

Related reading

• What is GEO? A plain-English guide for business owners
• Is your business showing up in ChatGPT? A 10-minute self-check
• How to write content AI actually quotes

About the author

Bantu Nonyukela is the founder of SearchLimelight. He has spent ten years in SEO, after starting out as a front-end developer, and is a Toptal-verified SEO expert. From Cape Town, he helps small and medium businesses in South Africa, the US, the UK and Australia get found in search and named by AI. Connect on LinkedIn

Sources

• Ahrefs, “An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)”: ahrefs.com
• Ahrefs research roundup, including the YouTube mentions and AI misinformation findings, May 2026: businesswire.com
• Ahrefs, analysis of AI Mode and AI Overviews citations: ahrefs.com
• SparkToro and Gumshoe, research on AI recommendation consistency, 2026: sparktoro.com
• Aggarwal et al., “GEO: Generative Engine Optimization,” KDD 2024: arxiv.org/abs/2311.09735
• BrightLocal, Local Consumer Review Survey 2026: brightlocal.com
• European Broadcasting Union and BBC, News Integrity in AI Assistants, October 2025: ebu.ch
• Seer Interactive, AI Overviews and CTR, September 2025 update, reported by Search Engine Land: searchengineland.com
• Pew Research Center, July 2025: pewresearch.org