Why It Matters
AI agents are becoming a primary way users discover and interact with content. ChatGPT, Claude, Perplexity, and Google's AI Overviews all need to read and understand your site. If your site isn't AI-ready, you're invisible to a growing segment of traffic.
The w2agent score is like a Lighthouse performance audit — but instead of measuring page speed and accessibility for browsers, it measures how well AI systems can access, understand, and represent your content.
The 7 Categories
w2agent evaluates websites across seven categories, each covering a different aspect of AI readiness:
Bot Accessibility
15 pointsCan AI crawlers actually reach your pages? This checks robots.txt, your sitemap, HTTP status codes, response times, and whether robots.txt blocks user-initiated AI agents (ChatGPT-User, Claude-User, Perplexity-User) or AI search fetchers (OAI-SearchBot, PerplexityBot). Training crawlers like GPTBot and ClaudeBot are reported but not scored.
Structured Data
15 pointsDoes your site include schema.org markup that helps AI classify content? Checks for Article, Organization, Product, FAQ, and BreadcrumbList schemas — the types AI models most frequently use.
Content Quality
15 pointsAre your pages substantive enough for AI to extract useful information? Checks meta descriptions, heading structure, content length, and whether pages have enough text to summarize.
Agent Discovery
12 pointsCan AI find your site's most important content? Checks for llms.txt (and whether its links resolve), an A2A agent card, and markdown alternates of your pages.
Agent Protocols
13 pointsDoes your site support emerging AI protocols? Checks for MCP endpoints, API endpoints, forms and action schemas agents can use, whether AI agents' user agents can reach your pages, and UCP on commerce sites.
Technical Foundations
10 pointsAre there technical barriers? Checks for client-side-only rendering, JavaScript-required content, excessive redirects, and slow response times.
Auth & Access
20 pointsCan an agent use your API on a user's behalf? Checks for a public API, OpenAPI spec, developer portal, OAuth discovery with PKCE, scoped permissions, and an RFC 9727 API catalog. Checks that depend on an API you don't offer are marked not applicable instead of failing.
Scoring
The overall score is 0-100, calculated as a weighted sum of all category scores. Each individual check within a category contributes a specific number of points. The final score maps to a letter grade:
| Grade | Score | Meaning |
|---|---|---|
| A | ≥ 75 | Excellent — your site is well-prepared for AI agents |
| B | ≥ 60 | Good — minor improvements possible |
| C | ≥ 45 | Fair — several areas need attention |
| D | ≥ 30 | Poor — significant gaps in AI readiness |
| F | < 30 | Failing — major barriers prevent AI access |
Grades use the unrounded score, while the report shows it rounded, so a site at 74.6 shows 75/100 with a B.
What "Good" Looks Like
A site scoring 75+ typically has:
- ✓ An llms.txt file at the site root
- ✓ robots.txt that doesn't block user-initiated AI agents or AI search fetchers
- ✓ Schema.org JSON-LD on key pages
- ✓ Descriptive meta descriptions on every page
- ✓ Server-rendered HTML (not client-side-only)
- ✓ An XML sitemap with all important pages
- ✓ Fast response times (<1.5s for AI crawlers, <500ms for full marks)
Beyond the Score
A score is just a starting point. The real value of an AI readiness audit is the specific, actionable recommendations — which files to create, which configurations to change, and which structured data to add. w2agent doesn't just score your site — it generates the files you need to fix the issues it finds.
Score Calculation Example
Here's how a real score breaks down for a typical marketing site that has llms.txt and schema.org but blocks AI search fetchers in robots.txt:
| Category | Max | Earned | Issue |
|---|---|---|---|
| Bot Accessibility | 15 | 10 | AI search fetchers blocked |
| Structured Data | 15 | 12 | Missing FAQPage schema |
| Content Quality | 15 | 14 | — |
| Agent Discovery | 12 | 7 | No agent-card.json |
| Agent Protocols | 13 | 4 | No MCP or API endpoints |
| Technical Foundations | 10 | 9 | — |
| Auth & Access | 20 | 14 | No OAuth discovery |
| Total | 100 | 70 | Grade B |
This site earns a B (70) — solid content and tech, but loses points on bot access rules and missing agent protocols. Adding agent-card.json and unblocking AI search fetchers would move it toward an A (75+) without changing any content.
Why Scores Differ by Site Type
The seven categories are weighted to reflect what matters most for AI access. Sites with heavy client-side rendering (SPAs, Next.js without SSR) often fail the Technical category even if everything else is correct. E-commerce sites frequently lose points on Structured Data because product pages lack complete Schema.org Product markup. Documentation sites tend to score highest because they're designed for reading — clean HTML, good headings, and lots of linkable content.
The biggest quick win for most sites is Agent Discovery: add llms.txt and an agent card, and the 12-point category moves from near 0 to ~10 with under an hour of work. The second biggest is Bot Accessibility — fixing common blocking issues costs nothing and can recover up to ~4 points immediately.
Related Articles
- What is llms.txt? — The fastest way to boost your Agent Discovery score.
- Why AI Crawlers Get Blocked — Fix Bot Accessibility issues that silently drop your score.
- Schema.org for AI — Structured data that improves your Structured Data category score.
Score your site now
Get your free w2agent score and generate the files your site needs.
Get Your Score