Let me skip the preamble.
Last week I ran 29 of the world's most recognized brands — Apple, Google, Microsoft, Amazon, OpenAI, you name it — through a systematic Generative Engine Optimization audit. The kind of audit that checks whether these brands can actually be found, cited, and recommended by AI systems like ChatGPT, Perplexity, and Google AI Overviews.
The result?
Every single brand scored zero on AI Discoverability. Not low. Zero.
OpenAI — the company that literally builds ChatGPT — scored 20 out of 100 on overall GEO health. Anthropic, the team behind Claude, got 30. If the people creating AI search can't pass an AI visibility check, we have a structural problem, not a company-specific one.
The Full Dataset
Here's every brand, scored across five dimensions: AI Discoverability, Content Structure, Competitor Gap, Source Authority, and Brand Consistency. Total score is out of 100.
| Brand | Industry | Total | AI Disc. | Content Struct. | Source Auth. | Brand Consist. | Issues |
|---|---|---|---|---|---|---|---|
| Shopify | E-commerce | 39 | 0 | 90 | 48 | 71 | 3 |
| HubSpot | Marketing | 39 | 0 | 90 | 48 | 71 | 3 |
| ByteDance | Technology | 39 | 0 | 90 | 48 | 71 | 3 |
| Semrush | Marketing | 39 | 0 | 90 | 48 | 71 | 3 |
| Notion | Technology | 39 | 0 | 90 | 48 | 71 | 3 |
| Apple | Technology | 30 | 0 | 65 | 43 | 61 | 4 |
| Technology | 30 | 0 | 65 | 43 | 61 | 4 | |
| Microsoft | Technology | 30 | 0 | 65 | 43 | 61 | 4 |
| Nike | Retail | 30 | 0 | 65 | 43 | 61 | 4 |
| Samsung | Technology | 30 | 0 | 65 | 43 | 61 | 4 |
| WebMD | Health | 30 | 0 | 65 | 43 | 61 | 4 |
| Alibaba | E-commerce | 30 | 0 | 65 | 43 | 61 | 4 |
| Anthropic | Technology | 30 | 0 | 65 | 43 | 61 | 4 |
| Ahrefs | Marketing | 30 | 0 | 65 | 43 | 61 | 4 |
| Moz | Marketing | 27 | 0 | 55 | 41 | 57 | 5 |
| Amazon | E-commerce | 25 | 0 | 50 | 40 | 55 | 5 |
| Netflix | Entertainment | 25 | 0 | 50 | 40 | 55 | 5 |
| Salesforce | Technology | 25 | 0 | 50 | 40 | 55 | 5 |
| Tencent | Technology | 25 | 0 | 50 | 40 | 55 | 4 |
| Xiaomi | Technology | 25 | 0 | 50 | 40 | 55 | 5 |
| Huawei | Technology | 25 | 0 | 50 | 40 | 55 | 5 |
| Adobe | Technology | 22 | 0 | 40 | 38 | 51 | 6 |
| Tesla | Automotive | 20 | 0 | 35 | 37 | 49 | 7 |
| Wikipedia | Education | 20 | 0 | 35 | 37 | 49 | 7 |
| OpenAI | Technology | 20 | 0 | 35 | 37 | 49 | 7 |
| JD.com | E-commerce | 20 | 0 | 35 | 37 | 49 | 7 |
| Stripe | Fintech | 20 | 0 | 35 | 37 | 49 | 7 |
| Zapier | Technology | 20 | 0 | 35 | 37 | 49 | 7 |
| Mayo Clinic | Health | 20 | 0 | 35 | 37 | 49 | 7 |
Average total score: 27.4 out of 100. Average AI Discoverability: 0.
Three Patterns That Actually Matter
After staring at this data for a few days, three patterns jumped out. They are not what the GEO industry is selling you.
Pattern 1: Content marketers beat brand marketers. Every time.
The top five — Shopify, HubSpot, ByteDance, Semrush, Notion — all scored 90 on Content Structure. These are not the biggest brands in the world. But they are the ones that invest most heavily in educational content, documentation, and structured knowledge bases.
Shopify does not just sell e-commerce tools. They publish guides, tutorials, case studies, and a massive help center — all with clean HTML structure, proper headings, and internal linking that makes their content machine-readable.
Compare that to Tesla (35 on Content Structure) or OpenAI (also 35). Tesla's website is basically a car configurator with a blog bolted on. OpenAI's site is a series of press releases and API docs with no coherent knowledge architecture.
Meanwhile, Mayo Clinic — one of the most trusted health information sources on the planet — also scored 35 on Content Structure. Their website is a labyrinth of JavaScript-rendered content, dynamic pages, and poor semantic markup. All that medical expertise, invisible to AI systems.
The pattern is clear: it is not about how much you know, it is about how well your content is structured for machines to read.
Pattern 2: SEO tool makers are quietly winning the GEO game
Semrush (39), Ahrefs (30), Moz (27) — the companies that have spent a decade teaching others about search optimization are themselves better prepared for AI search than most of their clients.
This is not a coincidence. These companies already have:
- Massive content libraries built on topic clusters
- Clean URL structures and internal linking
- Schema markup on key pages
- Teams that understand technical SEO deeply
The irony? Most of these companies have not explicitly done any "GEO optimization." They just kept doing good SEO with proper structure, and that turns out to be 80 percent of what GEO requires.
Pattern 3: The AI industry is hilariously bad at its own game
This is the one that made me laugh. OpenAI scored 20. Anthropic scored 30. The companies building the AI systems that will supposedly replace search engines cannot even get their own websites optimized for those same systems.
OpenAI's website has 7 critical issues flagged — missing llms.txt, no structured data, poor semantic markup, inconsistent brand signals. These are the people who built ChatGPT. If they cannot optimize their own site for AI discoverability, what makes you think your web designer down the street can?
Here is my honest take: the entire GEO services industry is selling a solution to a problem that even the AI companies have not solved. The playbook does not exist yet. Anyone claiming they have a "proven GEO methodology" is either lying or lucky.
What "AI Discoverability = 0" Actually Means
Let me be precise about this. AI Discoverability measures whether a brand gets cited or recommended when relevant queries are asked across AI platforms. A score of 0 does not mean the brand does not exist — it means when someone asks an AI assistant about that category, the brand does not show up in the response.
Ask ChatGPT "what is the best e-commerce platform?" It might mention Shopify (which scored highest overall at 39). Ask it "what is the best automotive brand?" It probably will not mention Tesla in a structured, citable way — because Tesla's website does not give AI systems the structured signals they need.
This is a fundamentally different game from traditional SEO. In SEO, you optimize for rankings. In GEO, you optimize for citation. You are not trying to be the first result — you are trying to be the answer.
The Four Things That Actually Move the Score
Across all 29 audits, the brands that scored higher consistently had these four things in common:
1. An llms.txt file in the root directory. This is a new standard — essentially a README for AI crawlers. It tells AI systems what your site is about, what topics you cover, and what you are an authority on. Most of the 29 brands do not have one. The ones that do (Shopify, HubSpot, Semrush) scored higher.
2. Schema.org structured data on key pages. JSON-LD markup that tells AI systems "this is an Organization, this is our name, this is what we do, here is our expertise." It is not magic — it is basic machine-readable metadata. But 80 percent of websites do not have it.
3. Topic-cluster content architecture. Not just blog posts — interconnected content hubs around specific topics. Shopify's e-commerce guides, HubSpot's marketing academy, Semrush's SEO learning center. These create what AI systems recognize as topical authority.
4. Consistent brand signals across platforms. Same name, same description, same expertise areas on your website, Wikipedia, Crunchbase, LinkedIn, and industry directories. AI systems cross-reference multiple sources to determine if you are a legitimate authority. Inconsistent signals mean lower trust.
What This Means for the GEO Industry
I have been writing about the GEO space for a while now. I have called out the scammers, the hype merchants, and the people selling snake oil. This data reinforces what I have been saying:
The GEO opportunity is real. The data proves that nobody — not Apple, not Google, not OpenAI — has cracked AI visibility yet. The playing field is genuinely level. A small company with great structured content can outperform a Fortune 500 brand in AI search.
But here is what the 4,800+ self-proclaimed "GEO service providers" do not want you to know: nobody has a proven, repeatable methodology for this. We are all figuring it out. The tools are new. The AI systems are changing weekly. What works today might not work when GPT-5 or Gemini 3.0 ships.
The brands that will win at GEO are the ones that focus on fundamentals:
- Structure your content for machines, not just humans
- Build genuine topical authority through consistent, deep content
- Make your brand identity machine-readable across every platform
- Stay adaptive — the AI landscape changes fast
Methodology
This audit was conducted using a systematic GEO evaluation framework that tests five dimensions:
- AI Discoverability (25%): Whether the brand is cited or recommended across major AI platforms (ChatGPT, Perplexity, Google AI Overview)
- Content Structure (25%): Technical readiness of website content for AI consumption — semantic HTML, structured data, llms.txt, crawlability
- Competitor Gap (20%): Relative positioning vs. direct competitors in AI citations
- Source Authority (15%): Brand trust signals across platforms — Wikipedia presence, industry citations, media mentions
- Brand Consistency (15%): Cross-platform identity coherence — same name, description, and expertise signals everywhere
All 29 brands were tested between August 8-11, 2026. Tests were automated and run from the same server location to control for geographic variation. AI platform responses may vary by region and over time.
The free audit tool used for this research is available at SilkGeo free GEO audit tool. You can run the same test on your own website.
The Bottom Line
If Apple, Google, and OpenAI cannot pass an AI visibility check, you should not feel bad about your own scores. But you should feel motivated.
The window is open. The big brands have not figured this out yet. The companies that invest in structured, AI-friendly content now — while the competition is still sleeping — will have an enormous advantage when AI search fully matures.
This is not about buying a tool or hiring a GEO agency. It is about understanding that the web is being re-indexed by a new generation of machines, and most of the internet is not ready.
Your move.
This report was produced by the SilkGeo research team. We build free GEO tools because we believe the industry needs more transparency and less hype. Try the free audit on your own site.