# GEO Audit

> Free Generative Engine Optimisation audit tool. Paste a URL and see how visible your site is to AI search engines like ChatGPT, Perplexity, Claude, and Google AI Overviews.

## What is GEO?

Generative Engine Optimisation (GEO) is the practice of making websites visible to AI-powered search engines. Traditional SEO focuses on ranking in Google's blue links — GEO focuses on being cited in AI-generated answers from systems like ChatGPT, Perplexity, Claude, and Google AI Overviews.

The term was formalised by researchers at Princeton and Georgia Tech (Aggarwal et al., 2023) who found that specific content optimisations — structured data, content depth, authoritative sourcing, and answer-first writing — significantly increase the likelihood of being cited in AI-generated responses.

## How GEO Audit Scores Your Site

We run 14 checks across three categories, each scored 0–10 and weighted to produce a final score out of 100.

### Technical (35%)
- **robots.txt AI rules** — Are AI crawlers (GPTBot, ClaudeBot, PerplexityBot) explicitly allowed?
- **llms.txt** — Do you serve a structured summary for language models at /llms.txt?
- **Schema coverage** — What percentage of pages have JSON-LD structured data, and how diverse are the types?
- **Sitemap** — Is there a discoverable sitemap referenced in robots.txt?
- **Meta descriptions** — Do pages have substantive titles and descriptions for retrieval scoring?
- **HTTPS** — Is the site served securely?
- **OG & Twitter completeness** — Are Open Graph and Twitter Card tags present across pages?

### Content (35%)
- **Content depth** — Median word count and thin page ratio. Deeper content gets cited more.
- **Heading quality** — Proper H1 usage and heading hierarchy for AI parsing.
- **Image accessibility** — Alt text coverage. Images without descriptions are invisible to AI.
- **Answer-first content** — Do pages lead with substantive answers rather than preamble?

### Authority (30%)
- **Entity clarity** — Organization schema, sameAs links, and consistent author information.
- **Agent discoverability** — RSS feeds, sitemaps, OpenAPI specs, and llms.txt as discovery mechanisms.
- **Trust signals** — Consistent identity, links to known platforms, and schema diversity (E-E-A-T).

Grades: Excellent (80+), Good (60–79), Needs Work (40–59), Not GEO-ready (<40).

## Sources

- [GEO: Generative Engine Optimization (Aggarwal et al., 2023)](https://arxiv.org/abs/2311.09735) — the foundational Princeton/Georgia Tech study
- [Google Structured Data Documentation](https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data)
- [OpenAI Crawlers Documentation](https://platform.openai.com/docs/bots)
- [Anthropic ClaudeBot Documentation](https://support.anthropic.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler)
- [llmstxt.org Specification](https://llmstxt.org/)
- [W3C WCAG 2.1](https://www.w3.org/TR/WCAG21/)
- [Open Graph Protocol](https://ogp.me/)
- [Twitter/X Card Markup](https://developer.x.com/en/docs/x-for-websites/cards/overview/markup)
- [Zyppy — How AI Overviews Shift Traffic](https://zyppy.com/seo/ai-overviews-to-google-ads/) — Cyrus Shepard's research on AI citation patterns

## API for AI Agents

GEO Audit has an API endpoint designed for AI agents and automated tools. No authentication required.

### Run an audit

```
GET https://geoaudit.co.uk/report?url=example.com
```

**Parameters:**
- `url` (required) — domain or full URL to audit (e.g. `example.com` or `https://example.com/page`)
- `fresh=true` (optional) — force a new scan even if a recent result exists
- `wait=true` (optional) — hold the connection open until the audit completes (up to ~280s). No polling needed — ideal for single-fetch AI agents like ChatGPT

**Response format:**
- Default: plain text (`text/plain`) containing Markdown-formatted content — readable by any browser or AI agent
- JSON: Send `Accept: application/json` header

**Status codes:**
- `200` — audit complete, response contains full Markdown report with scores, breakdown, and recommendations
- `202` — audit in progress or just started. Check the `Retry-After` header (10s for in-progress, 20s for just started) and retry the same URL
- `400` — missing or invalid `url` parameter

### Recommended: single-request mode (for AI agents)

```
GET https://geoaudit.co.uk/report?url=example.com&wait=true
```

The server holds the connection open and returns the full report when ready. If the audit takes longer than ~280 seconds, returns a timeout message with a link to check back.

### Alternative: polling workflow

1. `GET /report?url=example.com` → receive `202` with `Retry-After: 20`
2. Wait 20 seconds
3. `GET /report?url=example.com` → receive `202` with `Retry-After: 10` (still processing)
4. Wait 10 seconds
5. `GET /report?url=example.com` → receive `200` with full Markdown report

### Force a fresh scan

If the user requests a new scan of a previously audited domain:
```
GET https://geoaudit.co.uk/report?url=example.com&fresh=true
```

## Claude Code Skill

Developers using Claude Code can install a skill to run GEO audits directly from their editor:

```
/geo-audit example.com
```

Setup instructions and the full SKILL.md file: https://geoaudit.co.uk/integrate

The skill handles URL detection, API polling, result presentation, and offers to implement recommendations when working on the audited site's codebase.

## Links

- [Run a free audit](https://geoaudit.co.uk): Paste any URL and get an instant GEO score
- [Full methodology](https://geoaudit.co.uk/methodology): Detailed scoring breakdown with research citations
- [Agent API](https://geoaudit.co.uk/report?url=example.com): Direct API endpoint for AI agents
- [Integrate](https://geoaudit.co.uk/integrate): Claude Code skill, curl, and CI/CD usage