Your AI Visibility Tool Measures the Wrong Thing

On January 10, 2026, we published something that didn't exist before: the AI Visibility Definition. Version 1.1.0. A technical standard for how websites communicate with AI systems.

Here's why that matters more than you'd think.

The Terminology Problem Nobody's Solving

Ask three different agencies what "AI visibility" means. You'll get three completely different answers.

One says it's about getting mentioned in ChatGPT responses. Another insists it's SEO with extra steps. The third starts talking about tracking tools that charge £500/month to tell you whether Claude knows you exist.

They're all describing different things while using the same words.

That confusion costs businesses money. Real money. You buy a tool thinking it'll fix your AI visibility problem, except the tool measures something entirely different from what you actually need. The tool tracks outcomes. What you needed was infrastructure validation.

Until now, there was no authoritative source that said: "Here's what these terms actually mean. Here's how they relate to each other. Here's what you're paying for when someone says they'll improve your AI visibility."

6 Definitions Every Website Owner Needs in 2026

The AI Visibility Definition establishes precise technical definitions for six interconnected terms. Each has a term code (AV-001 through AV-006) for unambiguous reference.

AI Visibility (AV-001)

The degree to which a website can be discovered, correctly interpreted, accurately represented, and safely cited by AI systems including large language models, AI search engines, and retrieval-augmented generation systems.

This is the goal. Everything else serves it.

AI Visibility Checking (AV-002)

The process of technically verifying whether a website's infrastructure enables AI systems to discover, interpret, trust, and safely use that website as an information source.

This validates capability, not outcomes. It answers: "Can AI systems technically understand and trust this website?" Not: "Do AI systems currently mention this website?"

Critical distinction. We'll come back to it.

AI Discovery Files (AV-003)

Machine-readable files published to communicate information to AI systems, forming what's becoming the web's identity layer for AI. The specification defines ten core files:

• llms.txt - AI-readable business identity and context
• llm.txt - Compatibility variant (redirects to llms.txt)
• llms.html - Human-readable HTML version with Schema.org data
• ai.txt - AI usage permissions and intent signals
• ai.json - Machine-parseable AI interaction guidance
• identity.json - Structured canonical identity data
• brand.txt - Brand naming and representation rules
• faq-ai.txt - Factual Q&A formatted for AI consumption
• developer-ai.txt - Technical and platform context
• robots-ai.txt - AI crawler-specific access directives

Plus supporting files: robots.txt, sitemap.xml, security.txt, humans.txt, and structured data markup.

AI Visibility Tracking (AV-004)

The ongoing measurement of how a website or brand appears in AI-generated responses over time. This monitors mention frequency, sentiment, accuracy of representation, and changes in how AI systems describe you.

Tracking measures outcomes. It observes what AI systems currently do, not what they're technically capable of doing.

AI Visibility Monitoring (AV-005)

Continuous observation of AI system behaviour related to your brand, typically with automated alerts for changes. A subset of tracking focused on real-time observation.

AI Retrieval Testing (AV-006)

Querying AI systems with specific prompts to observe whether and how they reference your brand in their responses. Results are inherently variable, depending on prompt phrasing, model version, and temperature settings. Indicative, not deterministic.

The Critical Distinction: Checking vs Tracking

This is where most people, and most tools, get confused.

AI Visibility Checking validates infrastructure. It's deterministic and verifiable. Run it twice, get the same results. Anyone can inspect and validate the findings independently.

AI Visibility Tracking measures outcomes. It observes what AI systems are actually saying about you right now. Variable by nature. Run it twice, potentially different results. Because AI systems change, prompts vary, and sampling parameters shift.

Both matter. But they solve different problems.

"Large language models increasingly rely on website information, but face a critical limitation: context windows are too small to handle most websites in their entirety."

llms.txt specification, llmstxt.org

That's the core of why infrastructure matters first. If AI systems can't process your site efficiently, tracking how often they mention you is pointless. It's like buying a speedometer when your car won't start. The speedometer works fine. It's measuring the wrong problem.

If your AI visibility score is 23/100, fixing that requires infrastructure work: proper AI discovery files, consistent identity signals, structured data. That's what Checking validates.

If your infrastructure scores 95/100 but ChatGPT still describes you incorrectly, that's a Tracking problem. The infrastructure is fine; the AI's interpretation or training data is the issue.

Why 365i Published This Definition

We've been building AI visibility infrastructure since before most people realised it was a category.

The AI Visibility Checker, our free tool that's now analysed thousands of websites, needed a clear framework. When we tell you your score is 47/100, what does that actually mean? What are we measuring? And when we offer the AI Site Identity service for £295 + VAT, what exactly are we building? How does that relate to tracking tools? To prompt testing?

The definition answers those questions. Not just for our customers, but for the industry.

"If your content isn't in the Docs section of an llms.txt file, your brand and your narrative will not be found."

Jeremy Nguyen, Swinburne University academic and AI researcher

Nguyen's warning lands differently once you've got the terminology sorted. He's talking about AV-003 (AI Discovery Files) and their impact on AV-001 (AI Visibility). Without the files, the checking score drops, and the tracking results follow. The chain is clear when the vocabulary is precise.

That's why we licensed the definition under Creative Commons BY 4.0. Anyone can use it, adapt it, build on it. Proper citation required, but the goal is standardisation, not gatekeeping.

The Reference Implementation

Definitions need reference implementations. Something that says: "This is what AI Visibility Checking looks like when done correctly."

The 365i AI Visibility Checker is that reference implementation. It performs a read-only technical audit of your website's AI-facing infrastructure. Validates presence, accessibility, and consistency of AI discovery files. It doesn't query AI models or simulate prompt responses. It validates capability.

Free to use. Takes about 10 seconds. Gives you a score out of 100 with specific recommendations.

What This Means for Your Website

Three practical implications.

1. You can audit your tools

When someone sells you an "AI visibility" solution, you can now ask: "Which definition does this address? Are you checking infrastructure (AV-002)? Tracking outcomes (AV-004)? Running retrieval tests (AV-006)?"

If they can't answer clearly, they probably don't understand the problem they're claiming to solve.

2. You can prioritise correctly

Infrastructure first. Always. If your AI discovery files are broken or missing, tracking is pointless. You're measuring outcomes of a system that can't see you properly.

Check your infrastructure. Fix what's broken. Then track outcomes to see if AI systems interpret it correctly.

3. You can compare properly

The definition includes machine-readable formats in JSON and YAML. Tooling can be built against this. Comparisons become meaningful when everyone's using the same vocabulary.

What To Do Now

If you've read this far, you care about whether AI systems understand your business correctly. Three options:

1. Read the full definition at ai-visibility.org.uk. Technical where it needs to be, clear throughout.
2. Check your current status with the AI Visibility Checker. Free. Thirty seconds gives you a baseline.
3. Get it sorted with our AI Site Identity service: all ten core files, content audit, validation testing. £295 + VAT, 24-48 hours delivery.

The terminology problem is solved. The framework exists. As we covered in our post about AI files versus SEO, these are distinct disciplines. Now there's a standard to prove it.

Frequently Asked Questions