The GEO/AEO Audit
A single-paste audit that finds the gap between the pages Google ranks and the pages AI assistants actually cite, traces each problem to its cause, and hands you a prioritized fix list for your own site.
Published June 24, 2026 Updated June 24, 2026
Plain markdown. Paste it into any AI assistant and it runs the audit on your own site, phase by phase.
Already have it? View your other resources ->
An audit for both search surfaces at once.
Search split into two surfaces. There is the results page, where Google ranks ten links, and there is the answer surface, where an assistant reads the web and hands back one composed answer with a few named sources. Most audits only look at the first.
This one checks both at once and returns a prioritized fix list specific to your site: which pages to rewrite, which to protect, and which to cut. It is run by pasting a single markdown file into an AI assistant, which works through the analysis with you and asks for the handful of inputs it needs.
It is built to be honest about uncertainty. Every finding cites the evidence behind it, and the assistant is told to challenge its own conclusions before presenting them.
The pages Google keeps and the pages AI cites are not the same set.
An SEO audit asks one question: will Google rank this page. A GEO and AEO audit asks a second, separate one: when someone asks an assistant in your space, does your page get pulled into the answer, and do you get credited.
These two questions can give opposite answers about the very same page. A page can be too thin for Google to bother indexing while being the exact page an assistant quotes when it builds an answer. That split is the most valuable thing this audit surfaces.
The rule that falls out of it: never hide a page from search to fix a quality problem if that page is already earning AI citations. Hiding it removes it from the AI engines too. Such pages get rewritten, not removed.
Five phases, in order, plus a checklist.
After a short intake, the audit moves from diagnosis to causes to a ranked roadmap, then to the levers that actually move citations and the loop that tells you it worked.
The dual-surface diagnosis — What Google keeps vs. what AI cites
Cause-level analysis — Trace every problem to its cause, not its symptom
Triage and the prioritized roadmap — Keep and deepen, consolidate, or prune
Apply the levers, drop the myths — Spend effort where the research points
The re-measurement loop — A light monthly snapshot per engine
It also ships with a compact checklist you can run by eye between full audits. The whole thing is one plain markdown file. Enter your email above to get it, then paste it into your assistant.
The whole method in five lines.
If you only remember one thing from each phase, remember these. The downloadable file expands every one into concrete steps with the evidence behind each call.
The dual-surface diagnosis
Build two lists and compare them: the pages Google indexes and ranks, and the pages AI engines cite. The gap between what Google keeps and what AI cites is the core finding.
Cause-level analysis
For each problem find the cause, not the symptom, across nine dimensions: thin or templated content, AI-crawler reachability at the CDN edge, rendering, answer-first formatting, schema, sitemap and freshness, titles and meta, internal linking, and off-site authority.
Triage and the prioritized roadmap
Sort every issue by effort against impact, and bucket every weak page into keep-and-deepen, consolidate, or prune. Never hide a page that earns AI citations; rewrite it instead.
Apply the levers, drop the myths
Original statistics and first-hand data, inline citations, and unlinked brand mentions are what move citations. AI-specific files, schema-as-magic, and Core Web Vitals as a primary lever are not. The audit tells you both.
The re-measurement loop
Set up a light monthly snapshot of citations and grounding questions per engine, plus indexed versus crawled-not-indexed counts, so you can see whether the fixes actually moved anything. Changes take weeks, not days.
Built and run on my own site first.
I built this audit and ran it on my own site before writing it down. It caught something I almost got wrong: a block of programmatic pages Google had declined to index, labelled “crawled, currently not indexed,” were the exact same pages earning 536 Copilot citations in three months, holding 62.5 percent of the cited sources on one core question. The instinct was to hide the thin pages from search. That would have killed the single best distribution channel the site had. The audit put them on the rewrite list, not the delete list. The number is real and unembellished; that catch is what this audit is designed to make routine.
Anyone who wants to be the source engines rank and assistants cite.
Run a site, documentation, or a product, and care about both surfaces of search. Useful whether or not you have AI traffic yet.
You run a content or media site
You publish guides, reviews, or analysis and you want to be the source engines rank and assistants cite. The audit finds the pages worth protecting, the ones worth rewriting, and the ones quietly dragging the rest down.
You run documentation or a product site
When someone asks an assistant how to do something in your category, you want your docs to be the grounded, cited source. The same dual-surface check finds where the answer is buried or the crawler is blocked.
You run a personal or expert site
Founders, senior engineers, marketers, and writers who want to be the named source. Useful whether or not you have AI traffic yet: it finds the gaps that decide whether you get cited at all.
Questions before you start.
What is the difference between a GEO/AEO audit and an SEO audit?
An SEO audit asks one question: will Google rank this page. A GEO and AEO audit (generative engine optimization, answer engine optimization) asks a second, separate question: when someone asks an AI assistant in your space, does your page get pulled into the answer and credited. The same page can get opposite answers, and that gap is the whole point of this audit.
How do I run it? Do I need to install anything?
No install. It is a single plain-markdown file. Paste the whole file into Claude, ChatGPT, Gemini, or Claude Code and the assistant runs the audit with you, phase by phase. If you run it inside an agent with access to your repository and live site, it can trace problems to the code that causes them instead of guessing from symptoms.
What data do I need to provide?
Four inputs, and it works with whatever you have: a technical crawl (Ahrefs, Screaming Frog, or similar), your Google Search Console Pages data, your Bing Webmaster AI Performance data if you have any AI traffic, and a link to the latest official Google guidance on AI search. Missing one does not block the audit; it just flags the gap.
Will it tell me to delete my thin pages?
Not blindly. The core rule of this audit is to never hide a page from search if it is already earning AI citations, because hiding it removes it from the AI engines too. Those pages get rewritten, not removed. The audit sorts every weak page into keep-and-deepen, consolidate, or prune, which are three different actions that look identical from the outside.
Do I need AI traffic already for this to be useful?
No. It is useful whether or not you have AI traffic yet. If you do, the Bing AI Performance data shows exactly which pages get cited and on which questions. If you do not, the audit still finds the thin content, blocked crawlers, and answer-format gaps that decide whether you get cited at all.
The diagnosis, then the work.
This audit tells you where the gaps are. These playbooks are how you close them.
How to Get Cited by AI ->
The offense that follows this diagnosis: the six-phase GEO playbook for earning citations from ChatGPT, Copilot, and Perplexity once the audit shows you where to push.
SEO Strategy Playbook ->
The other half of the funnel: the nine-phase system that earns the ranking on the traditional results page while the GEO work earns the citation.
Title & Meta Audit Workflow ->
A focused fix for one cause this audit surfaces: titles and metas that rank but do not convert, scored against a rubric and measured at Day 7, 14, and 30.
Find the gap before it costs you.
One paste, both surfaces, a ranked fix list. Get the audit and run it on your own site.
Plain markdown. Paste it into any AI assistant and it runs the audit on your own site.
Already have it? View your other resources ->