> ## Documentation Index
> Fetch the complete documentation index at: https://docs.koalr.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# GEO Audit
> Generative Engine Optimisation: how well your pages are written for AI platforms to extract, quote and cite, with prioritised fixes.
AI can only cite what it can read and understand. The GEO Audit crawls your pages and scores them for AI readiness, then hands you the fixes in priority order.
## Site GEO Score
The dial at the top is your site-wide AI readiness, 0 to 100, averaged across the pages Koalr has scored. Beside it: **Pages Scored**, **Citations** (how many times AI has cited pages from your site across all tracked prompts), **With Issues**, and a **Grade Distribution** of your pages.
Three coverage rings show how much of the site has **schema markup** (structured data that helps AI understand a page), **FAQ markup** (question-and-answer structure is highly citable) and **fresh content** (updated in the last six months; AI platforms favour recent pages).
## The pages table
Every crawled page with its type, score and grade, open issues, its signals (schema and FAQ markup), how often AI has cited it, and its freshness (Fresh, Aging, Stale, Outdated), grouped by score band from Poor upwards. Paste any URL into **Add a page URL to audit** and choose **Scan Page** to score it on demand; **Re-scan** refreshes every page. If a scan is interrupted, a banner says so and offers **Try again**, and your existing results stay.
Click a page for its [detail view](/analyse/geo-audit-page-detail).
## How scores are calculated
Each page is scored across seven categories drawn from published Generative Engine Optimization research:
| Category | What it checks |
| -------------------------- | ------------------------------------------------------------------------------ |
| **Content Extractability** | Can an AI lift a clean answer out of the page without the surrounding noise |
| **Structure** | Headings, lists and sections that map to the questions people ask |
| **Answerability** | Does the page answer a question directly, early, in a form that can be quoted |
| **Entity Clarity** | Is it obvious who you are, what the product is, and what it is called |
| **Grounding Signals** | Specifics: numbers, names, dates, sources, the things AI uses to trust a claim |
| **Authority Context** | Signals that the page is a credible source in its category |
| **Readability** | Plain sentences, short passages, no walls of text |
Pages under about 500 words rarely get cited and are flagged as too thin to score reliably. **Passage Quotability** highlights the sections most likely to be lifted whole.
## Marking pages not optimised
Some pages are not meant to be found in AI: checkout, thank-you, login and internal pages. Tick them and choose **Mark not optimised**. Three things happen:
1. They disappear from the Action Centre.
2. Kevin stops recommending improvements to them.
3. They stop counting towards the Site GEO Score, the grade distribution, the coverage rings and the headline counts.
They stay in the table in a collapsed **Not optimised** section at the bottom and can be restored at any time with **Restore to audit**. This is for pages you have decided not to optimise, not for pages that are simply scoring badly.
## Kevin's Top Tips and the Action Centre
The highest-priority findings for the site appear at the top of the page. The full list, page by page, lives in the [Action Centre](/act/action-centre) under On-site, with each finding linked back to the page and the check that raised it.
Do the GEO Audit alongside [Agent Experience](/analyse/agent-experience). The audit measures how citable a page is once read; Agent Experience measures whether AI crawlers are allowed to read it at all. A perfect page behind a blocked crawler scores nothing where it matters.