The AI Visibility Index: We Tested 1,004 Businesses Across 5 AI Models. 66% Are Completely Invisible.

When a potential customer opens ChatGPT and asks "what are the best project management tools," ChatGPT gives them an answer. That answer contains 4 or 5 brand names. If your brand isn’t in that list, you didn’t lose the sale — you were never considered. You didn’t even get a chance to lose.

That’s the new layer of discovery sitting on top of Google. And we wanted to know how many businesses were actually in it.

So we ran the largest public study of AI visibility we could build. 1,004 businesses. 10 industries. 5 AI models. 95,392 individual data points. Every business was enriched with 13 measurable signals — Domain Authority, Google Review Count, schema markup, llms.txt presence, blog content volume, and more. Then we ran Pearson correlations and logistic regression to figure out what actually moves the needle.

The headline finding is uncomfortable:

> 66% of the businesses we tested received zero mentions across all five AI models.

Not low visibility. Not mentioned once. Zero. No version of any prompt in any model produced any reference to these businesses. They don’t exist in the AI recommendation layer. At all.

"Two-thirds of the businesses we tested aren’t losing to competitors on AI — they don’t exist at all," says Joel House, founder of MentionLayer and author of *AI for Revenue*. "There’s no version of the response that mentions them. No sentence. No link. No acknowledgment. They’re not in the conversation, and most of them don’t even know it’s happening."

This article walks through the full findings, the statistical analysis behind each one, a self-assessment so you can score your own business, and a tiered 90-day action plan based on where you currently are. The full anonymized dataset is at the bottom if you want to run your own analysis.

Let’s start with the ten findings you need to know.

If you only have a minute, read this section. Every finding below is covered in full later in the article with the supporting data.

1. 66% of businesses are completely invisible. Across 1,004 businesses tested, 660 received zero mentions across all 5 AI models. Only 4% appeared in all five.

2. Domain Authority and Google Review Count are the twin pillars. DA correlates with AI visibility at r=0.337. Google Review Count correlates at r=0.333. They are virtually tied, and together they explain more variance than every technical signal combined.

3. Review count matters. Review rating barely does. Google Review Count: r=0.333. Google Review Rating: r=0.056. AI models care about *volume* as a proxy for brand prominence, not whether you have 4.2 or 4.8 stars.

4. There’s a "1,000 Review Cliff." Visibility hovers around 34–39% from 0 to 500 reviews. At 1,000+ reviews it jumps to 54%. At 5,000+ it hits 77%.

5. AI models fundamentally disagree. Only 11% of mentioned businesses appear in 2+ models. If one model mentions you, there’s an 89% chance the next one won’t. You can’t optimize for "AI" as a monolith.

6. llms.txt is a mid-market play, not a silver bullet. Raw correlation of r=0.141 disappears almost entirely when you control for Domain Authority. It helps businesses in the DA 21–60 range (+14–15 points). For DA below 20, no benefit. For DA above 60, not needed.

7. Blog quality beats page quantity. Blog post count: r=0.145 positive. Total indexed pages: r=-0.020 negative. More targeted content helps. More pages does not.

8. Blocking AI crawlers has zero measurable impact. Pearson r=0.009. The training data horse already left the barn.

9. 77% of AI mentions are positive. 0.2% are negative. There is no such thing as bad AI press. AI models either recommend you or pretend you don’t exist.

10. Industry variation is enormous. Personal Injury Law: 71% visible. Med Spas: 17% visible. The gap between the most and least visible industry is 54 points.

Now let’s look at each finding in detail, with the data, the why, and what it means for your business.

Before the findings, here’s the short version of the methodology. The full technical details are in the appendix at the bottom.

Sample: 1,004 businesses across 10 industries. Local industries (law, real estate, dental, med spa, accounting, plumbing) were all LA-based. National industries (SaaS CRM, SaaS PM, financial advisors, digital marketing agencies) sampled the most prominent players.

Models tested: ChatGPT (GPT-4o), Perplexity (sonar-pro), Gemini (2.5-flash), Claude (Sonnet), Google AI Overview (via SerpApi).

Prompts: 20 unique buying-intent prompts per industry across 6 categories (direct recommendation, comparison, specific need, conversational, authority-seeking, decision).

Detection: Two-stage. Heuristic string matching first (exact name, partial name, domain), then Claude Sonnet at temperature 0 verified every match to filter false positives.

Enrichment per business: Domain Authority (Moz API), Google Reviews (count and rating), robots.txt policy, schema markup, SSR check, citability score, blog post count, total indexed pages, llms.txt presence, llms-full.txt presence, FAQ content, comparison content, sitemap check.

Statistical methods: Pearson correlations (n=248 to n=1,000), logistic regression with L2 regularization, segment profiling, controlled variable analysis.

Total data points: 95,392.

Full methodology, caveats, and download link are in the appendix.

The headline stat is brutal, and it needs to be unpacked.

Only 34% of businesses were mentioned by at least one AI model. Just 4% appeared in all five. The average business in our sample appears in 0.8 models out of 5. The *median* business appears in zero.

Look at the gap between Invisible and Highly Visible. Visible businesses have roughly 2x the Domain Authority, 4–6x the review count, and 3–4x the blog content. These aren’t subtle differences — they’re order-of-magnitude gaps.

But here’s the part that’s easy to miss: the Invisible segment isn’t made up of tiny businesses with no online presence. The average invisible business in our sample has 555 Google reviews and 128 blog posts. These are functioning, established businesses. They’re visible in Google. They run ads. They have customers. And AI has no idea they exist.

### The Opportunity Frame

"When 66% of your category is completely invisible to AI, that isn’t a threat — it’s the cheapest arbitrage window I’ve seen in fifteen years of digital marketing," says Joel House. "The brands that move in the next 12 months will lock in AI associations that become nearly impossible to displace later."

The important strategic insight here is that invisibility is the default state, not the exception. If two-thirds of your competitors are also invisible, then you’re not behind — you’re at the starting line. The businesses that understand this and act in the next 12–18 months will establish positions that are dramatically cheaper to claim now than they will be once the category wakes up.

Of the 13 variables we measured, two stood head and shoulders above the rest:

• Domain Authority: r = 0.337 (n=1,000)
• Google Review Count: r = 0.333 (n=901)

These two signals are virtually tied as the strongest single predictors of AI visibility. Together, they explain more variance than every other signal combined.

Here’s the full correlation table, sorted by strength:

"Every SEO tool is trying to sell you a button to push," says House. "The data says there is no button. The two things that predict AI visibility are the two things that take the longest to build — domain authority and genuine customer volume. Nothing beats them, and nothing replaces them."

Three things jump out of this table that deserve their own callout.

### Review Count Matters. Review Rating Doesn’t.

Google Review Count correlates at 0.333. Google Review Rating correlates at just 0.056 — barely above noise. AI models care about volume, not your star rating. A business with 2,000 reviews at 4.1 stars is dramatically more visible than one with 15 reviews at 5.0 stars. Volume is the signal of brand prominence. Rating is treated as secondary.

This has a direct practical implication: stop chasing review perfection and start chasing review volume. The hour you spend trying to get a 3-star review removed is an hour you could have spent generating three new 4-star reviews. The data is clear about which one moves your AI visibility.

### FAQ Content and Schema Markup Are Near-Zero

Despite being heavily recommended by SEO tools, FAQ content (r=0.036) and schema markup (r=0.072) show negligible correlation with AI visibility. They’re not harmful, but they’re not the lever. If you’ve been prioritizing schema rollouts as your AI strategy, the data says reallocate.

### More Pages Is *Slightly Negative*

Total indexed pages correlates at -0.020 — essentially zero, but directionally negative. **Businesses with thousands of thin pages are slightly *less* visible than those with targeted content.** Quality over quantity isn’t a platitude here. It’s what the data shows.

When two signals are both strong, the next question is: if I can only invest in one, which matters more?

We ran a quadrant analysis to find out:

The finding is unambiguous. High DA with low reviews (37%) beats low DA with high reviews (26%). A business with strong domain authority and weak review presence is more visible to AI than a business with weak domain authority and strong review presence.

"If you can only invest in one thing, build domain authority," says House. "A business with DA 50 and 30 reviews is more visible to AI than one with DA 15 and 3,000 reviews. That’s the opposite of what most ‘AI SEO’ advice will tell you, but it’s what the data shows every time we run the numbers."

The combination (48%) is where the real visibility lives. Getting both right roughly doubles your visibility vs. having either signal alone while the other is weak.

### The Isolation Test

To confirm this, we looked at extreme scenarios where only one signal was present:

Domain Authority alone gets you almost halfway. 47% of businesses with DA 40+ are visible even with minimal reviews and no technical optimization. That’s striking.

Reviews alone barely help without DA. Businesses with 500+ reviews but DA under 20 are only 10% visible — just 5 points above the absolute baseline of 5%.

The baseline for businesses with nothing is 5%. If you have low DA, few reviews, and no technical optimization, you have a 1-in-20 chance of appearing in any AI model. That’s where 660 of the 1,004 businesses in our study live.

Review count doesn’t create a smooth gradient of AI visibility. It creates a cliff.

From 0 to 500 reviews, visibility barely moves: 34% to 39%. A 5-point gain for 10x the review effort. Barely worth chasing.

But cross 1,000 reviews and visibility jumps to 54% — a 15-point leap. The average number of models mentioning the business goes from 1.1 to 1.6. Getting from 500 to 1,000 reviews produces three times the marginal visibility gain as getting from 0 to 500.

We call this the 1,000 Review Cliff, and it’s one of the most actionable findings in the study.

"The jump at 1,000 reviews isn’t magic," says House. "It’s the point where your review count stops looking like ‘this business runs a review campaign’ and starts looking like ‘this business has real customer volume.’ That’s the signal the models are picking up on."

Any business can generate 200–500 reviews through a systematic ask campaign. Getting to 1,000+ typically requires years of operation, real customer volume, and sustained satisfaction. It’s a signal that’s hard to fake. And because it’s hard to fake, AI models appear to weight it heavily as a proxy for genuine prominence.

### The Practical Implication

If you’re below 1,000 reviews, getting above that threshold is probably the single highest-leverage investment you can make in AI visibility this year. It’s more valuable than any technical optimization. More valuable than any content project. More valuable than any schema rollout.

If you’re already above 1,000, the returns continue but diminish. Going from 1,000 to 2,000 adds 12 points of visibility. Going from 2,000 to 5,000 adds 11 more. The cliff is the biggest single jump on the entire curve.

Domain Authority creates a similar pattern — not a smooth slope, but a staircase with distinct steps:

Three clear steps emerge:

Step 1: DA 0–20 — The Invisibility Zone. Below DA 20, you’re fighting a 78–92% invisibility rate. You are not getting recommended by AI at this level, no matter what technical tricks you try.

Step 2: DA 21–50 — The Plateau. Notice that DA 21–30 (35%), DA 31–40 (37%), and DA 41–50 (35%) are essentially flat. The staircase *pauses* in the mid-range. This is a critical insight: once you’re past the invisibility zone, pure DA gains stop producing visibility gains. Something else has to kick in.

That something else is the combination of reviews, content, and press that we cover in other findings. At this plateau, adding more backlinks without also building review volume and content depth produces diminishing returns.

Step 3: DA 51+ — The Acceleration Zone. Above DA 51, each tier adds 10–20 points of visibility. At DA 81+, you’re in the 64% visibility range — effectively the opposite of the baseline. These are businesses AI models recognize as authoritative almost by default.

### What This Means by Stage

Your DA range tells you what your next move is:

• DA 0–20: Your job is to escape the invisibility zone. Focus on link building fundamentals — digital PR, guest posting, resource page outreach. You need to cross DA 20 before anything else matters.
• DA 21–50: You’re on the plateau. More links alone won’t help. This is the stage where reviews, content depth, and llms.txt actually start to contribute meaningfully (see Finding 6).
• DA 51+: Every additional tier is worth chasing. You’re in the acceleration zone where DA gains translate directly into visibility gains. Invest in high-authority press and industry publications.

One of the most strategically important findings in the study is also the least talked about: AI models fundamentally disagree on who to recommend.

• Only 11% of mentioned businesses appear in 2 or more models
• 0% of businesses in our sample appeared at the raw mention level across all 5 models equally
• If one model mentions you, there is an 89% chance the next model won’t

This is not a unified "AI recommendation layer." It’s five different worldviews.

"You can’t optimize for ‘AI’ as one thing," says House. "ChatGPT, Perplexity, Gemini, and Claude disagree with each other 89% of the time on who to recommend. They’re not a single recommendation layer — they’re five different audiences with different biases, and you need a different strategy for each."

### Model-by-Model Mention Rates

Perplexity is the most generous recommender at 11.1% — 6x higher than ChatGPT. When Perplexity mentions you, 87.9% of the time it’s a positive recommendation. Perplexity’s real-time web retrieval means it’s pulling from current sources, making it the most responsive to fresh changes in your online presence.

ChatGPT is binary. At 1.6%, ChatGPT is selective — but when it mentions you, it’s *always* a recommendation (100%). ChatGPT doesn’t hedge. It either knows you well enough to recommend you, or it doesn’t mention you at all.

Gemini mentions nobody. Our API testing (2.5-flash) returned 0%. This likely reflects differences between API behavior and the consumer chat interface. We flag it as a methodological caveat — the public Gemini experience may be different.

Claude is the most independent. Claude had the least overlap with any other model in our cross-model analysis.

### The Model Overlap Matrix

This table answers: of the businesses mentioned by the Row model, what percentage are ALSO mentioned by the Column model?

The most important cell in this matrix: Google AIO → Perplexity: 84%. If Google’s AI Overview mentions you, there’s an 84% chance Perplexity will too. These two share the most similar worldview, likely because both rely heavily on real-time web retrieval.

ChatGPT → Gemini: 70% suggests these parametric models (relying on training data) share ground on which brands are "in the training set."

Claude is the outlier. Claude’s overlap with other models is consistently the lowest (32–63%), suggesting it weights different signals or has a distinct training data composition.

### The Two-Track Strategy

This finding means you need a two-track strategy, not a one-size-fits-all approach.

Track 1 — Real-Time Models (Perplexity + Google AIO): These reward fresh web signals. Recent Reddit threads, current review profiles, fresh press coverage, new blog content. If your strategy is to move AI visibility in 30–60 days, this is the track you optimize for.

Track 2 — Parametric Models (ChatGPT + Gemini): These reward deep historical authority. Long-standing brand recognition, heavy training data presence, Wikipedia entries, and years of consistent signals. This is a 6–12 month track.

The businesses that score highest across all models have *both*. They generate fresh signals continuously AND they have years of historical authority. That’s how you get into all five models.

This is the finding that will generate the most pushback, so let’s be precise about what the data shows.

Raw correlation: llms.txt presence correlates with AI visibility at r=0.141 (n=248). That ranks #5 out of 13 variables. On the surface, it looks like a modest but real positive signal.

But here’s the problem: businesses that adopt llms.txt are systematically different from those that don’t.

Businesses with llms.txt have 617% more blog posts, 59% higher schema scores, and are dramatically more likely to have sitemaps and schema markup. They’re not average businesses who added a file. They’re the businesses already investing heavily in everything.

Much of the "llms.txt effect" is simply measuring "this business does SEO well."

### Controlling for Domain Authority

To isolate the actual independent effect of llms.txt, we controlled for Domain Authority and ran the analysis within buckets:

For small businesses (DA <20), llms.txt provides zero benefit. These businesses don’t have enough underlying authority for the file to matter. AI models don’t know who they are, and an llms.txt file doesn’t change that.

For mid-market businesses (DA 21–60), llms.txt shows a genuine +14–15 point lift. This is statistically meaningful and practically significant.

For large brands (DA 61+), llms.txt shows no meaningful benefit. Sample sizes are small (n=4 to n=40), but these brands appear to be so well-known that the file is irrelevant.

### The Honest Take

"Every SEO tool in the market right now is selling llms.txt as the AI visibility silver bullet," says House. "We tested it against 1,004 businesses. The honest answer is it only helps if you’re in the middle — DA 21 to 60. If you’re small, it doesn’t do anything. If you’re already big, you don’t need it. The industry is selling a mid-market play as a universal fix."

llms.txt is worth adding if your DA is between 21 and 60. The data supports a genuine 14–15 point visibility boost in that range. If your DA is below 20, spend your time building domain authority first — the file won’t compensate for the lack of underlying signal. If your DA is above 60, you probably don’t need it, and the opportunity cost of engineering time is better spent elsewhere.

The SEO industry’s current obsession with llms.txt is a classic case of survivorship bias. The businesses loudly reporting their llms.txt success are the same businesses that already had strong SEO fundamentals. They’d likely be visible to AI with or without the file.

Two content signals point in opposite directions:

• Blog post count: r = 0.145 — moderate positive
• Total indexed pages: r = −0.020 — slightly negative

More quality blog content correlates positively with AI visibility. More total pages does not, and may very slightly hurt.

This is the clearest quality-over-quantity signal in the entire study. Businesses with hundreds of targeted, high-quality blog posts are more visible than businesses with thousands of thin or auto-generated pages.

### Technical SEO Combinations

We also tested how combinations of technical signals perform:

The full technical stack (llms.txt + Schema + Sitemap) at 71% visibility looks impressive, but remember the confounding problem — businesses running the full stack are the same ones with high DA and extensive content.

The more telling comparison: Reviews 1,000+ without llms.txt (53%) vs the full technical stack (71%). The gap is real but not huge, and reviews alone — with zero technical optimization — still get you more than halfway there. Authority signals outperform technical signals every time.

### The Practical Rule

Write fewer, better blog posts. Do not flood your site with AI-generated content thinking volume will help. It won’t. A site with 80 genuinely useful posts will outperform a site with 800 mediocre ones. AI models aren’t impressed by large sitemaps — they’re influenced by content that gets cited and referenced in the threads, forums, and articles AI models actually read.

One of the cleanest findings in the study: blocking AI crawlers via robots.txt has no measurable effect on AI visibility.

• Pearson r = 0.009 (essentially zero)
• DA correlation with blocking: r = −0.117 (higher DA businesses are slightly *less* likely to block)

Why? Because these businesses were already in the training data before they started blocking crawlers. ChatGPT, Claude, and Gemini were trained on data from before most businesses implemented AI-specific crawler blocks. The horse already left the barn.

This doesn’t mean blocking is futile forever. As models shift toward more real-time retrieval (like Perplexity already does), blocking may eventually reduce visibility. But as of April 2026, in our data, it’s a non-factor.

If you’re currently blocking AI crawlers out of principle, that’s fine — it’s not costing you visibility right now. If you’re blocking them hoping it will protect your content from being trained on, you’re closing the barn door years after the horse left.

AI models are overwhelmingly positive when they mention a business:

• 77% positive (recommendation, praise, favorable context)
• 23% neutral (listed without strong sentiment)
• 0.2% negative

Out of 95,392 data points, we found a single-digit number of negative mentions. There is essentially no such thing as "bad AI press."

"Here’s a reframe most marketers haven’t processed yet: there is no such thing as bad AI press," says House. "The models don’t write negative reviews. They either recommend you or they pretend you don’t exist. The entire risk equation shifts from reputation management to presence management."

This is fundamentally different from traditional search, where negative reviews or hit pieces can rank for your brand name. In the AI recommendation layer, the binary is presence vs. absence, not positive vs. negative.

The strategic implication is important: the only bad AI visibility outcome is invisibility. You don’t need to worry about AI models "saying bad things" about your brand. You need to worry about them saying nothing.

This also means the old playbook of "defensive SEO" — suppressing negative results, owning your brand SERP — doesn’t translate. The defensive play in AI visibility isn’t suppressing negatives. It’s generating presence.

AI visibility varies enormously by industry. The gap between the most and least visible is 54 percentage points.

### The Top 10 Most Visible Businesses Overall

SaaS dominates the top of the leaderboard. The exception is The Dominguez Firm, a local law firm competing with global SaaS brands — the result of decades of aggressive investment in reputation, media, and digital authority.

"If you run a med spa, a dental practice, or a home services business, your industry is structurally disadvantaged for AI visibility," says House. "The entire category has low domain authority and local-only presence. The good news is that means your competition is also invisible — the first mover in each local market gets a near-monopoly."

### What to Do Based on Your Industry

If you’re in SaaS (CRM, PM, MarTech): You’re competing in the most visible category in our study. The bar is high — your competitors have DA 70+, thousands of G2/Capterra reviews, and comparison content written about them constantly. Your lever is review volume on G2/Capterra/TrustRadius and comparison content that ranks for "X vs Y" queries. Do not bother with generic blog SEO; target comparison and alternative keywords.

If you’re in Personal Injury Law or Legal Services: You’re in the most visible industry (71% visible in LA). The winners are firms with decades of history, heavy local press, and large review volumes. Your lever is case result publicity, local press relationships, and aggressively building toward the 1,000 review cliff.

If you’re in Real Estate: 61% of LA realtors are invisible. The visible ones are almost all individuals with strong personal brands (Aaron Kirman, agents who dominate local press). This is an agent-brand-building play, not a brokerage play. Invest in personal PR, podcast appearances, and Zillow review volume.

If you’re in Dental, Med Spa, or Home Services: Your industry is structurally hard — 74–83% invisible. But that’s also the opportunity. The bar to beat your local competition is low. Focus on Google Review volume (this is achievable locally), local press, and entity consistency across directories. The first business in each local market to systematically pursue AI visibility wins a near-monopoly because nobody else is trying.

If you’re in Financial Advisory: 69% invisible. The winners are national names (Vanguard, Fidelity) with massive authority. If you’re a RIA or local advisor, you’re not going to out-brand Vanguard — your lever is content authority on specific niches (retirement planning for teachers, business exit planning, etc.) plus local press.

If you’re a Digital Marketing Agency: 65% invisible. You’re competing with WebFX, Ignite Visibility, and other scaled agencies. Your lever is founder-led content, industry publications (Search Engine Land, MarketingProfs, Forbes Agency Council), and podcast presence — thought leadership is the primary signal for agencies.

If you’re in Accounting: 60% invisible. The winners are Big Four (EY, etc.). For smaller firms, niche vertical specialization plus high review volume in your local market is the path.

Before reading the action plan, spend five minutes scoring your own business. The numbers below are calibrated against the data from this study — they’ll give you a rough AI Visibility Score and tell you which action plan tier you’re in.

Assign yourself points in each category, then add them up.

### 1. Domain Authority (max 25 points)

Check your DA using Moz, Ahrefs DR, or any free DA checker.

• DA 0–10: 2 points
• DA 11–20: 6 points
• DA 21–30: 10 points
• DA 31–40: 13 points
• DA 41–50: 15 points
• DA 51–60: 18 points
• DA 61–70: 21 points
• DA 71–80: 23 points
• DA 81+: 25 points

### 2. Google Review Count (max 25 points)

• 0–100 reviews: 3 points
• 101–300: 6 points
• 301–500: 9 points
• 501–999: 12 points
• 1,000–1,999: 18 points *(you crossed the cliff)*
• 2,000–4,999: 22 points
• 5,000+: 25 points

### 3. AI Model Presence (max 20 points)

Open ChatGPT, Perplexity, Gemini, and Claude. Ask each one: "What are the best [your category] in [your location or for your use case]?" Count how many mention your brand by name.

• 0 models: 0 points
• 1 model: 5 points
• 2 models: 10 points
• 3 models: 15 points
• 4–5 models: 20 points

### 4. Blog Content (max 10 points)

• 0–25 quality blog posts: 1 point
• 26–100 posts: 4 points
• 101–300 posts: 7 points
• 300+ posts: 10 points

*(Thin auto-generated content doesn’t count. We’re talking about posts a human would actually read.)*

### 5. Press & Third-Party Mentions (max 10 points)

In the last 12 months, how many third-party publications have mentioned your brand (news articles, guest posts, podcast interviews, award lists, industry publications)?

• 0–2: 1 point
• 3–9: 4 points
• 10–24: 7 points
• 25+: 10 points

### 6. Entity Consistency (max 10 points)

Search Google for your exact brand name. Check the top 10 results.

• Inconsistent or missing presence across major directories: 2 points
• Present but with inconsistent descriptions/categories: 5 points
• Consistent across all major platforms (Google Business, LinkedIn, Crunchbase, industry directories): 8 points
• All of the above PLUS a Google Knowledge Panel: 10 points

### Your AI Visibility Score

Add up your points. You’re in one of four tiers:

Write your score down. Now match it to the action plan below.

This is the part that separates this article from every other "AI visibility" post you’ve read. The action plan depends entirely on your current stage. Doing the wrong thing for your tier wastes months.

"The businesses winning at AI visibility aren’t doing anything clever," says House. "They’re not hacking prompts. They’re not gaming schema. They’re the same businesses that built real authority the old-fashioned way. AI models are rewarding the same thing Google has always rewarded — they’re just rewarding it louder."

Find your tier below and follow the plan in order.

---

### Tier 1: Invisible (Score 0–25)

Your situation: You are the median business in the study. DA is probably below 20, reviews are probably under 300, and zero AI models mention you. Baseline visibility is 5%.

Your strategic priority: Escape the invisibility zone. Don’t bother with technical tricks. They won’t work at this level.

Days 1–30 — Foundations: 1. Fix your entity data first. Claim and complete Google Business Profile, LinkedIn, and every major directory in your industry. Ensure name, address, website, category, and description match across all of them. This is a 10-hour project that unlocks everything else. 2. Start a systematic review request campaign. Every customer transaction should trigger a review request. Goal: get from wherever you are to 300+ Google reviews within 90 days. Use SMS + email. Your target is the 1,000 review cliff, but progress toward it starts now. 3. Audit your existing backlinks. Disavow spam. Identify any unearned authority you can reclaim.

Days 31–60 — Authority building: 1. Start a digital PR campaign. Target 3–5 industry publications or local news outlets per month. HARO/Qwoted responses, expert commentary, local business stories. The goal is earning DA 20+ within 6 months. 2. Publish 2–3 genuinely high-quality blog posts per month. Not AI slop. Real content that could be referenced by journalists, other bloggers, or forum users. 3. Begin community presence on Reddit/Quora in the subreddits and topic areas relevant to your category. Genuine contribution only. This plants the seeds for the citation layer.

Days 61–90 — Measurement: 1. Run the 60-second test monthly. Track your score. Document which models mention you, which mention competitors, which mention nobody. 2. Do NOT add llms.txt yet. It won’t help at your DA level. Wait until you cross DA 20.

What success looks like in 90 days: DA crosses 20. 100–200 new Google reviews. Entity data is clean. You’re mentioned in 1–2 local or industry publications. Your AI Visibility Score moves from 0–25 to 25–40.

---

### Tier 2: Emerging (Score 26–50)

Your situation: You have foundations. DA is probably in the 20–40 range. You have some reviews, some content, some presence. But you’re on the plateau — more of the same isn’t moving your AI visibility. You need to change what you’re investing in.

Your strategic priority: Break through the plateau. This is where reviews, content depth, and llms.txt start producing real returns.

Days 1–30 — Unlock the review cliff: 1. Make the 1,000 review goal your #1 priority. This is the single highest-leverage move at your stage. Build a review generation system: automated requests, follow-ups, incentive-free but friction-free. Target +50 reviews per month minimum. 2. Add llms.txt and llms-full.txt to your site. Your DA range is where the data shows a genuine +14–15 point lift. Do this week 1. 3. Publish a "best X" or "X vs Y" comparison post for your core category. This is the content AI models actually reference.

Days 31–60 — Citation seeding: 1. Identify the Reddit, Quora, and forum threads that AI models already cite when users ask about your category. Use Perplexity to find them — it shows sources. Target 20–30 high-authority threads. 2. Place genuine, helpful contributions in those threads. Your brand gets mentioned naturally, not as an ad. This is the fastest way to move Perplexity + Google AIO visibility. 3. Build authority content: 1–2 deep resources (2,000+ words) that journalists and bloggers would actually cite.

Days 61–90 — Press push: 1. Launch a press campaign targeting 5–10 industry publications. Founder story, unique data, customer case studies. 2. Track your share of model weekly across all 4 major models. Document changes. 3. Fix any remaining entity inconsistencies uncovered during the work.

What success looks like in 90 days: DA moves from mid-30s to mid-40s. Review count crosses 500, on track for 1,000 within 6 months. You appear in at least 2 AI models for at least one core prompt. Your AI Visibility Score moves from 26–50 to 45–65.

---

### Tier 3: Present (Score 51–75)

Your situation: You’re already in the 34% that gets mentioned. DA is probably 40–60. You have 500–2,000 reviews. Some AI models recognize you. Now the question is how to dominate rather than just show up.

Your strategic priority: Expand coverage across models and move from "mentioned" to "recommended."

Days 1–30 — Gap analysis: 1. Map which models mention you and which don’t. Perplexity + Google AIO are easier wins (real-time retrieval). ChatGPT + Gemini are longer plays (training data). 2. Analyze the top 10 businesses that beat you in AI recommendations. What threads do they appear in that you don’t? What publications cover them that don’t cover you? Build a target list. 3. Cross the 1,000 review cliff if you haven’t already. This single milestone is worth 10–15 points of visibility.

Days 31–60 — Citation dominance: 1. Scale citation seeding to 50–100 high-authority threads. The businesses at this tier invest heavily in being the name that comes up in forum discussions. 2. Launch comparison content campaign — "X vs Y" and "alternatives to Z" articles that rank for every major competitor. This is how you get into ChatGPT’s training for your category. 3. Secure 2–3 tier-1 press placements (Forbes, Inc, Entrepreneur, or your industry equivalents).

Days 61–90 — Authority compounding: 1. Pursue a Wikipedia entry if you’re notable enough. This is one of the highest-leverage AI training signals available. 2. Build out founder thought leadership — podcast appearances, industry speaking, founder LinkedIn content. 3. Start monitoring competitor AI visibility weekly. When a competitor gains visibility, figure out what triggered it and consider replicating.

What success looks like in 90 days: You appear in 3+ AI models consistently. You’re the recommended answer for at least one core category prompt. Competitor share of model is shifting measurably toward you. Score moves from 51–75 to 70–85.

---

### Tier 4: Dominant (Score 76–100)

Your situation: You’re in the top 13%. Multiple models mention you. You’re probably recommended, not just listed. Your challenge is maintaining dominance as competitors invest.

Your strategic priority: Lock in your position. Make displacement expensive.

Days 1–30 — Defense: 1. Run monthly share-of-model tracking across every major prompt in your category. Detect visibility drops within weeks, not months. 2. Identify the specific sources AI models cite when recommending you. Protect and expand those sources. 3. Audit your review sentiment trend. Volume is secured but declining sentiment can eventually affect visibility.

Days 31–60 — Moat expansion: 1. Publish original research. The kind that gets cited by every industry publication. This is how category leaders stay category leaders. 2. Launch a data-driven newsletter or report that other people reference. Becoming a source *for* AI models is higher leverage than optimizing to be found by them. 3. Maintain aggressive press presence — 1–2 top-tier placements per month.

Days 61–90 — Offensive moves: 1. Displace competitors in adjacent categories. You own your core — now extend into the categories immediately adjacent where you’re not yet the default answer. 2. Build the knowledge graph presence — Wikipedia, Wikidata, Crunchbase, industry encyclopedias. 3. Create the category vocabulary. The businesses that name the category (not just occupy it) become the reference point in AI training data.

What success looks like: Consistent presence across all 5 models. Recommended (not just listed) in majority of category prompts. Competitors would need to invest 6–12 months minimum to displace you.

---

Regardless of your tier, the one thing everyone should do is measure. Run the 60-second test monthly. Same prompts, same models, same day of the month. That’s your feedback loop.

"The reason most agencies can’t help clients with AI visibility is that they can’t measure it," says House. "You can’t improve what you don’t track. Run the 60-second test every month, across four models, with the same prompts. Watch the number move. That’s the only honest feedback loop in this space."

A study like this is only as useful as its honesty about limitations. Here’s what you should know before citing these numbers.

1. Correlation ≠ causation. DA and reviews correlate with visibility, but all three may be driven by underlying brand strength and age. Businesses don’t become visible *because of* high DA — they’re visible because they’re authoritative, and DA measures the same underlying thing. The action plan above still works because you can’t raise DA without raising the underlying authority it measures, but be careful about how you frame the mechanism.

2. Los Angeles market bias. Local industries in our sample are all LA-based — one of the most competitive markets globally. Results may differ in smaller markets, where the competition threshold for AI visibility could be lower and the opportunity window correspondingly larger.

3. Temporal snapshot. AI model outputs change frequently. This study was conducted in April 2026. Repeating it in 6 months may produce different absolute numbers, though we expect the *relative* importance of signals (DA > reviews > everything else) to remain stable.

4. Gemini’s 0% mention rate. Likely reflects API behavior differences versus the consumer Gemini chat interface. The API version (2.5-flash) may be more conservative than the interface users interact with. Treat Gemini numbers with caution until we can replicate with multiple Gemini API configurations.

5. Detection limitations. String matching may miss indirect mentions or misspellings. The AI-enhanced verification step catches many of these but not all. Our numbers are likely conservative — actual mention rates may be slightly higher than reported.

6. llms.txt sample size. Only 248 businesses had llms.txt data collected (those with verified own-domain URLs). Controlled analyses within DA buckets have small sample sizes (n=4 to n=40). Treat the directional patterns as credible but the exact percentage lifts as approximate.

7. Confounding in technical signals. Businesses that adopt llms.txt, schema markup, and other technical optimizations are systematically more sophisticated in every dimension. Raw correlations for any individual technical signal likely overstate its independent effect.

8. Industry composition. Our 10 industries are not a representative sample of the US economy. We chose industries that span local/national, high-authority/low-authority, B2B/B2C, and services/SaaS to provide a useful spread. Industries we didn’t test (healthcare, ecommerce, hospitality, manufacturing) may behave differently.

### Cross-Correlations (The Confounding Web)

The most important confound: DA and Google Review Count correlate at 0.268. Some portion of the reviews’ predictive power is actually measuring the same underlying construct as DA: brand prominence. This is why the action plan treats them as complementary rather than independent.

We’re publishing the full anonymized dataset so the industry can build on this research:

• [anonymized-dataset.csv](/data/ai-visibility-index-2026.csv) — 1,004 rows with all enrichment variables and AI visibility scores (business names hashed)
• [top-performers.csv](/data/ai-visibility-index-top-performers-2026.csv) — Top 5 businesses per industry with names, scores, and model-by-model visibility

If you run your own analysis and find something interesting, reach us at [email protected] or tag us on LinkedIn.

Citation format: If you’re referencing this study in an article, report, or presentation, please cite as: *House, J. (2026). The AI Visibility Index: A Study of 1,004 Businesses Across 5 AI Models. MentionLayer. https://mentionlayer.com/blog/ai-visibility-index-study*

Press inquiries: For press, podcast appearances, or speaking engagements related to this research, contact [email protected].

Here’s the uncomfortable truth buried in this data. AI visibility isn’t an optimization problem. It’s a land grab.

Right now, 66% of businesses are invisible. Most of them don’t know it and aren’t trying to fix it. The signals that move the needle (domain authority, review volume, press coverage, entity consistency) all take months to build. The businesses that start now will have positions locked in by the time their competitors realize the game has changed.

And the game has changed. 37% of consumers are already starting searches with AI. That number is moving up, not down. The brands that appear in AI recommendations aren’t just getting discovery — they’re getting discovery with the AI’s endorsement attached to it. That’s a different thing than a Google blue link. It’s closer to a friend recommending you.

"We’re in the fifteen-minute window where AI visibility is cheap to win and expensive to ignore," says Joel House. "In two years, this study will look like the SEO reports from 2005 that told people to claim their Google Business Profile. Obvious in hindsight. Impossible to catch up on if you waited."

The action plan in this article will get most businesses from invisible to present in 90 days and from present to dominant in 6–12 months. The businesses that do the work will own their categories. The ones that don’t will wonder where their customers went.

You now know more about what actually drives AI visibility than 99% of the marketing industry. You have a self-assessment. You have a tiered action plan. You have the full dataset to verify any claim in this article.

The only question left is whether you’re going to do anything with it.

If you want help running the audit, building the citation presence, or tracking your share of model, MentionLayer is the platform we built for exactly this. The 5-pillar audit is a good starting point — it establishes your baseline and generates a prioritized action plan specific to your business. Or just start with the 60-second test and move from there.

Whatever you do, start now. The window won’t stay open much longer.