AI Visibility Scoring: How to Calculate Whether Your Brand Actually Exists in the AI Answer Layer

AI visibility is whether AI engines cite your brand when someone asks a question you should own. Not whether your site ranks on page one. Whether ChatGPT, Perplexity, Claude, and Gemini pull your name, your data, and your claims into the answer. A SearchScore audit of 850,000+ websites found that 74.2% score as invisible or low visibility to AI search. Only 0.2% qualify as AI-ready. That is roughly 1,730 sites out of 850,000. If you cannot measure where you stand, you cannot fix it.

I have spent the last two years building systems that track exactly this: which brands AI engines extract, which they ignore, and what separates the two groups. The measurement problem is now the bottleneck. Most founders I talk to know AI visibility matters. Almost none can tell me their score.

What AI Visibility Actually Measures

Traditional SEO visibility counts rankings and clicks. AI visibility measures something fundamentally different: whether your brand appears inside an AI-generated answer at all.

The distinction matters because the answer layer operates on different rules. Google shows ten blue links. ChatGPT shows one synthesized response, sometimes citing three or four sources, sometimes citing none. Lumar's scoring methodology breaks it into two components: citation quality (25% weight) and brand mention quality (75% weight), both multiplied by your appearance rate across multiple prompt runs.

That formula reveals something most founders miss. A brand that appears in every run with a quality score of 80 will outrank a brand that appears in 10% of runs with a perfect 100. Consistency beats brilliance. Sporadic excellence is, mechanically, worse than showing up reliably.

Presenc AI's cross-industry study of 2,847 brands quantifies six weighted factors: Knowledge Presence (20%), Semantic Authority (18%), Citation Frequency (18%), Contextual Integrity (16%), Share of Voice (14%), and Recommendation Rate (14%). The composite score runs 0 to 100 across ChatGPT, Claude, Gemini, Perplexity, and Copilot, with minimum 90-day tracking windows. The cross-industry median is 49 out of 100, but the spread between top and bottom industries is 32 points.

The stakes are not abstract. Averi AI's B2B SaaS analysis found that AI-referred visitors convert at 14.2%, compared to 2.8% for Google organic traffic. AI-referred traffic to the top 1,000 websites grew 357% year over year. Yet only 18% of brands have an active AI visibility strategy. The gap between the opportunity and the response is enormous.

This is the first thing to internalize: AI visibility is not a single test. It is a pattern across hundreds of prompts, and the pattern rewards presence over perfection.

The 850,000-Site Benchmark: Where Most Brands Actually Stand

SearchScore's Q2 2026 AI Search Visibility Index is the largest public audit available. The numbers are worth sitting with.

Tier	Share of Sites	Approximate Count
AI-Ready	0.2%	~1,730
Strong	3.8%	~32,900
Emerging	21.8%	~188,900
Low Visibility	41.3%	~357,800
Invisible	32.9%	~285,000

The average score across all 850,000 sites: 34 out of 100. Down from 41.4 in March. The decline is not because sites got worse. SearchScore expanded its dataset 2.5x into the web's long tail, and the long tail is almost entirely invisible to AI engines.

Here is what I find most useful about their category breakdown. Technical foundation scores highest at 70.1 out of 100. On-page structure scores lowest at 23.1. The gap between the best technical score and the overall AI visibility average is 36 points. Roughly 90% of audited sites lack Organization schema markup, which SearchScore identified as the highest-leverage action for AI readiness, requiring only hours to implement. Most sites have the infrastructure to be found. They do not have the content architecture to be cited.

Industry Benchmarks: SaaS Leads, Construction Trails by 32 Points

Presenc AI's industry-by-industry breakdown reveals where the gaps fall hardest:

Industry	Median AI Visibility Score
SaaS/Technology	63
Cybersecurity	60
Media/Publishing	59
Financial Services	56
Education/EdTech	54
Healthcare/Pharma	49
Professional Services	45
Manufacturing/Industrial	37
Construction/Real Estate	31

The 32-point gap between SaaS (63) and Construction (31) is not random. SaaS companies produce documentation, publish benchmarks, and generate the kind of structured, data-rich content that AI engines extract easily. Construction firms run on project portfolios and trade references that AI engines cannot parse. If your industry trails the median, the opportunity is larger: you are competing against a field that has barely started.

Presenc AI also found that moving from the 40th to the 60th percentile takes 3 to 6 months of sustained effort. Advancing from the 75th to the 90th percentile requires 9 to 18 months. This is not a quick fix. It is an infrastructure investment.

The B2B Concentration Problem: Top 3 Capture 78%

The Linksii State of AI Search Visibility 2026 benchmark tracked 200 brands across 268 queries on ChatGPT, Claude, Gemini, and Perplexity, analyzing 7,278 citations across 1,350 domains. Their finding is stark: the top 3 brands in any category capture 78% of AI visibility. The remaining 197 brands split the other 22%.

That concentration is worse than traditional search. In Google, you might rank eighth and still get clicks. In AI answers, you are either in the response or you are not. There is no page two.

Linksii found that 104 of 200 tracked brands only appear when AI engines are asked about them directly. They never surface in category queries ("best CRM for agencies"), comparison queries, or problem-framed queries ("my sales pipeline is a mess"). The DerivateX B2B SaaS benchmark confirms the pattern at a finer grain: across 50 B2B SaaS companies, the gap between the highest scorer (Clio at 89) and the lowest (LeadSquared at 2) is 87 points.

There is also a 71% visibility drop when users shift from category queries to problem-framed queries. Brands that show up for "best project management software" often vanish when someone asks "how do I stop losing deals because my team can't coordinate." The problem queries are where buyers actually make decisions. And most brands are invisible there.

How AI Visibility Scores Are Calculated

The scoring frameworks vary across platforms, but the core mechanics converge on the same inputs. Here is what the major systems evaluate, distilled from SearchScore's 130+ signal methodology, Lumar's scoring architecture, and Presenc AI's six-factor composite:

Citation quality measures whether AI engines reference your content as a source. Not just whether they mention your brand, but whether they link to your pages, attribute data to your research, and pull direct quotes from your writing. GEO Tracker AI calls this "Citation Source Intelligence" and breaks it into three layers: extraction (recording the URLs cited), enrichment (classifying the source type and actionability), and action (generating appropriate responses for each venue type).

Brand mention quality measures how AI engines talk about you when they do mention you. Is the context positive, neutral, or negative? Is it accurate? Does it reflect your actual positioning or a competitor's framing?

Appearance rate is the consistency multiplier. How often does your brand appear across repeated prompts on topics you should own? This is where most brands fail. They might show up once for a specific query. They do not show up reliably across the query cluster. Linksii found less than 50% consistency in top brand recommendations across the four major AI platforms: what ChatGPT recommends, Claude may ignore entirely.

Topical authority reflects whether you own a coherent subject area or scatter content across unrelated topics. SearchScore found this category averaging 51.5 out of 100, meaning most sites have some topical signal but not enough depth to become the go-to source.

Content extractability is the structural factor: can an AI engine pull a clean, direct answer from your content? WhyIQ's AI Citability Index found that 44.2% of LLM citation extractions come from the first 30% of body text. Pages that bury the answer under 400 words of introduction score worse than pages that lead with the claim and then prove it. Their research also found that named-expert quotations produce a 28% citation lift, and that brand mentions correlate with AI visibility at r=0.334 to r=0.664, while backlinks correlate at just r=0.218. The old SEO playbook is mechanically less relevant.

The Consistency Multiplier Most Founders Ignore

I keep coming back to Lumar's formula because it exposes the mistake I see most often.

The visibility score uses the square root of your appearance rate as a multiplier. If you appear in 100% of runs, the multiplier is 1.0. If you appear in 25% of runs, it drops to 0.5. At 4% of runs, it is 0.2.

That square root softens the penalty compared to a linear model, but the effect is still dramatic. A brand with strong content that only surfaces in a quarter of relevant prompts loses half its potential score. Not because the content is bad. Because it is not consistently findable.

This is why one great blog post does not create AI visibility. You need a body of work that covers the query cluster so that AI engines encounter your brand repeatedly across different prompts in the same domain. Linksii's citation source analysis found that the top 20 domains generate 50% of all AI citations, led by Google's own properties (24.9%), YouTube (5.6%), Reddit (4.7%), and Wikipedia (3.4%). Reddit's share varies wildly by platform: 10% of Claude citations, 9% of ChatGPT, and 0% on Gemini and Perplexity.

A single definitive article helps. Twenty articles across the entity chain of related concepts, each citing primary data, each linking to each other, each structured for extraction: that is what pushes the appearance rate toward 1.0.

What 94% of AI Citations Have in Common

The earned media data tells the other half of the story. Research analyzing more than one million AI-generated links across ChatGPT, Claude, Gemini, and Perplexity found that non-paid media sources represent 94% of all AI citations. Earned media alone accounts for 25% of those citations. Muck Rack's generative search analysis, cited by Stacker, puts the number even higher: 95% of AI-generated citations come from unpaid media sources, with 85% of those originating from earned media specifically.

The composition of what gets cited matters. Press releases that AI engines cited contained roughly twice as many statistics as non-cited releases, 30% more action verbs, and 2.5 times more bullet points. The cited releases also had a 30% higher rate of objective sentences, meaning fewer superlatives and more verifiable claims.

One finding I cannot stop thinking about: the journalists most frequently pitched by PR professionals and those most frequently cited by AI engines share an average overlap of just 2%. Two percent. The entire traditional media-targeting model is optimized for a distribution network that AI engines barely use.

SparkToro's research, as reported by Spin Sucks, found that branded web mentions are the number one correlation with AI visibility. Not backlinks. Not domain authority. Mentions. Rand Fishkin put it bluntly: "Public relations is the future of marketing." The brands that score highest in AI visibility are the ones producing primary-source content: original research, first-party data, named frameworks, and specific operational claims that AI engines can extract and attribute.

How to Run Your Own AI Visibility Audit

Here is the audit I run for every brand I work with. You do not need a paid platform to start. You need discipline and about two hours.

Step 1: Map your query cluster. List the 15 to 20 queries a buyer would ask before choosing a solution in your category. Not your brand name. The problems they are solving. For AuthorityTech, that list includes queries like "how to get cited by AI search engines," "ai visibility measurement," and "earned media vs paid for AI citations."

Step 2: Run each query across four engines. ChatGPT, Perplexity, Claude, and Gemini. Record three things for each: whether your brand was mentioned, whether your content was cited with a link, and what the AI engine said about you.

Step 3: Calculate your appearance rate. Divide the number of queries where you appeared by the total queries tested. If you ran 20 queries across 4 engines (80 total prompts) and appeared in 12, your appearance rate is 15%. The square root of 0.15 is 0.387. That is your consistency multiplier. You are losing 61% of your potential score to inconsistency alone.

Step 4: Score citation quality. For each appearance, rate whether the AI engine linked to your content (high), mentioned your brand without linking (medium), or referenced your concept without attribution (low). Weight these 3, 2, and 1 respectively. Average across all appearances. GEO Tracker AI's three-layer framework offers a more granular model: it classifies each citation by source type, actionability state (Live, Limited, Frozen, Manual), and platform-specific signal strength.

Step 5: Identify the gaps. The queries where you did not appear are more valuable than the ones where you did. Each gap is a piece of content that does not exist yet, or exists but is not structured for extraction. Stacker's earned media analysis calls this the early mover advantage: initial AI associations become "sticky," and competitive displacement requires matching coverage and overcoming algorithmic historical weighting.

This manual audit is not precise. It is directional. And directional is enough to know whether you are in the 0.2% or the 74.2%.

Why Content Optimization Alone Cannot Fix the Score

The SearchScore data shows a 36-point gap between technical foundation (70.1 average) and overall AI visibility (34 average). Most websites have the technical basics in place. They still score terribly.

The missing layer is source authority. AI engines do not cite content because it is well-structured. They cite content because other trusted sources reference it, because it contains original data that cannot be found elsewhere, and because the entity behind it has been validated across multiple independent contexts. As Jonny Bentwood wrote in PRovoke Media, we are now in an "AI Reputation Economy" where AI systems actively form and share opinions about organizations based on the source signals they ingest.

This is where Machine Relations becomes the operational framework, not the marketing message. The discipline of building your brand's citation architecture across earned media, original research, and owned content so that AI engines encounter your authority from multiple independent directions. Structure gets you crawled. Authority gets you cited.

The GlobeNewsWire research confirms this mechanically. The highest citation rates occur within the first seven days of a release, and more than half of all citations reference material published within the prior 11 months. Sustained publishing cadence outperforms episodic campaigns. Zenodo's AIVI framework formalizes this into a quantifiable index for measuring entity presence in generative information engines, which confirms that the measurement infrastructure is maturing fast.

The brands getting cited are not the ones with the best single piece. They are the ones producing citable material consistently, across multiple formats, grounded in first-party evidence.

The Forcing Function: Score It or Stay Invisible

Here is where this lands. You either know your AI visibility score or you are guessing about whether your brand exists in the fastest-growing discovery channel in B2B.

The data says 74.2% of the web is invisible to AI engines. The data says the top 3 brands in any category capture 78% of visibility, leaving 22% for the remaining 197 competitors. The data says consistency beats brilliance by a measurable, formula-driven margin. And the data says 94% of AI citations come from non-paid sources, which means you cannot buy your way into the answer layer.

You can measure it today. Run the 80-prompt audit. Calculate the appearance rate. Identify the gaps. Or wait until the brands who already did this are so far ahead that the gap becomes permanent.

The answer layer is already answering questions about your category. The only question is whether your brand is in that answer.

FAQ

What is AI visibility?

AI visibility is the measure of how often and how prominently AI search engines (ChatGPT, Perplexity, Claude, Gemini) cite, mention, or reference your brand when users ask questions in your category. It is distinct from traditional SEO visibility, which measures rankings and clicks on search engine results pages. Presenc AI's cross-industry study found the median score across 2,847 brands is 49 out of 100.

How is an AI visibility score calculated?

Most scoring frameworks combine citation quality (whether AI engines link to and attribute your content), brand mention quality (how AI engines characterize your brand), and appearance rate (how consistently you show up across repeated prompts). Lumar's methodology weights brand mentions at 75% and citations at 25%, with a square-root consistency multiplier applied to both. Presenc AI uses a six-factor composite: Knowledge Presence, Semantic Authority, Citation Frequency, Contextual Integrity, Share of Voice, and Recommendation Rate.

What percentage of websites are visible to AI search engines?

According to SearchScore's Q2 2026 audit of 850,000+ websites, only 4% score as Strong or AI-Ready. 74.2% score as Invisible or Low Visibility. The average AI visibility score across all sites is 34 out of 100.

Does paid media improve AI visibility?

Research across more than one million AI-generated links found that non-paid media sources represent 94% of all AI citations. Muck Rack's generative search analysis puts it at 95%. Paid placement does not directly improve how AI engines cite your brand. Earned media, original research, and structured primary-source content drive AI visibility.

How long does it take to improve an AI visibility score?

According to Presenc AI's benchmark data, moving from the 40th to the 60th percentile takes 3 to 6 months of sustained effort. Advancing from the 75th to the 90th percentile requires 9 to 18 months. Stacker's analysis of earned media stickiness also found that initial AI associations become persistent, making early investment disproportionately valuable.

How can I check my brand's AI visibility right now?

Run 15 to 20 category-relevant queries across ChatGPT, Perplexity, Claude, and Gemini. Track whether your brand appears, whether it is cited with a link, and what context the AI engine provides. Divide appearances by total prompts to calculate your appearance rate. GEO Tracker AI's Citation Source Intelligence framework provides a more granular three-layer model for classifying each citation. The manual audit takes about two hours and gives you a directional score.