What is a GEO audit?
The featured-snippet answer, in plain language.
A Illucrum GEO audit is a 52-point manual review across seven sections: AI crawler and retrieval access, the classic SEO foundations that gate it, passage and chunk structure, entity and authority, structured data, prompt visibility testing across seven AI surfaces, and measurement and monitoring. It is delivered as a written report with every finding scored and sorted into four action tiers, alongside the full prompt log and the list of sources the engines actually cite in your category. There is one tier of audit. Delivery runs 5 to 8 business days.
The five stages of a GEO audit
Every audit follows the same five stages from intake to delivery. The order matters: run the prompts before you understand the category and you measure language nobody actually uses.
-
Intake (Day 1)
A short brief, written or over a 15-minute call (your choice). The intake covers:
· The site's URL and current state
· Your category, and the three competitors you'd name if asked
· The questions buyers ask you before they decide
· Any business context the data won't surface (a recent rebrand, a discontinued product, a merger that confused the record)
The intake matters more in GEO than in SEO, because the prompt set is built from it. Get the category and the buyer language wrong and every number that follows measures something nobody actually searches for.
-
Prompt set and crawl (Days 1 to 3)
We build the prompt set and agree it with you before anything runs. It covers:
· Category prompts: best tool, best provider, best option for a given use case
· Comparison prompts: you against each named competitor, and each competitor against the field
· Problem-led prompts: how buyers describe the pain before they know the category name
· Brand prompts: what the engines say about you unprompted, and whether it is correct
Each prompt is then run across five engines, repeatedly, while the site is crawled in parallel:
· ChatGPT, Perplexity, Gemini, Claude and Copilot, plus Google AI Overviews and AI Mode as separate surfaces
· Logged out, personalisation and memory off, two to three runs per prompt
· Crawl, rendering and edge checks for content parity and what AI agents actually receive
Every answer is logged: prompt, engine, date, result, sources cited. The log ships with the report so you can re-run any claim in it, and the prompt set becomes your permanent baseline.
-
Manual review (Days 3 to 5)
The 52-point checklist is applied by hand across seven sections. This is where most of the audit's value comes from. A few examples of what the manual stage catches:
· A CDN bot rule silently returning 403 to PerplexityBot while robots.txt looks perfectly correct
· A training-crawler block scoped broadly enough that it also removed the retrieval bots that produce citations
· Priority pages whose sections make no sense read in isolation, because every one opens with a back-reference
· An Article headline in schema that does not match the visible H1, on every page of a template
· A competitor's citations all originating from one roundup article you could be added to in a week
No tool currently reports why an engine skipped you. Monitoring tools produce a mention count, which is a symptom, not a diagnosis.
-
Scoring and prioritisation (Days 5 to 8)
Every finding is scored on two axes, then sorted into four tiers:
· Tier 1, fix immediately: high impact, low effort. Retrieval crawler blocks, edge blocking, missing author schema, headline mismatches. Usually hours of work.
· Tier 2, this quarter: high impact, medium effort. Answer-first blocks on priority pages, FAQ schema, date hygiene, entity markup completion.
· Tier 3, next quarter: high impact, high effort. Structural page rewrites, third-party presence building, publishing original data.
· Tier 4, de-prioritise: the things not worth your time, named explicitly so you stop spending on them.
The scoring rubric is consistent across every audit. Two-thirds of the value comes from tier 4: knowing what to ignore.
-
Report and handoff (Days 5 to 8)
The final report is delivered as a single PDF containing everything. It includes:
· A brief overview of the current situation, covering the main gaps and opportunities
· The full 52-point scoring table across all seven sections
· The complete prompt log, plus the cited source target list ranked by frequency
· The action plan in four tiers, with implementation notes for the top findings
If you want a 1h walkthrough call after delivery, that's included at no extra cost. If you don't, the report is written to stand on its own.
Timeline expectations
From the day you confirm the audit to the day the report lands, expect five to eight business days. The exact length depends on site size and how large the prompt set needs to be.
| Stage | Window |
|---|---|
| Intake and prompt set | Day 1 |
| Engine runs and crawl | Days 1 to 3 |
| Manual review | Days 3 to 5 |
| Scoring and report | Days 5 to 8 |
What you actually receive
The deliverable is concrete. Here's exactly what's in the PDF report at the end of an Illucrum GEO audit. Everything below sits inside one PDF. An optional 1h walkthrough call after delivery is included if you want it.
-
Overview
A summary of where you currently stand in AI answers, and the main gaps and opportunities.
-
Top issues found
The most important causes identified, ranked by how much of the visibility gap each one explains.
-
Detailed analysis by section
Each of the seven sections scored, point by point, with every finding explained in plain language.
-
The full prompt log
Every query, engine, date, answer and cited source, plus the ranked list of domains the engines draw on in your category.
-
A tiered action plan
Every finding sorted into fix-now, this quarter, next quarter, and de-prioritise.
-
Optional 1h walkthrough call
If you need it, we can schedule a call after delivery.
What a GEO audit isn't
The audit is a diagnosis, not a treatment. To set expectations honestly.
-
01
No guaranteed placements
There are no positions in an AI answer to buy or rank for. What you get is a diagnosis, a ranked plan, and a baseline. Anyone promising a placement is describing something that does not exist.
-
02
No monthly retainer
You don't sign up for ongoing work to get the audit.
-
03
No 200-page PDF dumps
Audits do not have to be long. They need to be useful.
If you decide you want help with implementation after the audit, that conversation happens after delivery, not before. See what the audit covers.
Process FAQs
The questions that come up most often about how the audit runs.
Do I need to give you access to GSC and Analytics?
Read-only access to Google Search Console and Google Analytics is strongly preferred, though the GEO audit depends on it less than an SEO audit would. Most of the evidence comes from the engines themselves. Where access helps is the measurement section: identifying existing AI referral traffic and setting up tracking you can keep using. Access is granted to a dedicated audit email and revoked at delivery.
Do you need our server logs?
If you can share them, yes. Server logs are the only reliable way to see which AI crawlers are actually reaching you and which are being turned away at the edge. Robots.txt tells you what you asked for; the logs tell you what happened. If logs aren't available we test from the outside instead, which catches most but not all of it.
Can we see the prompt set before you run it?
Yes, and you should. The prompt set is agreed with you at the end of intake. If it doesn't match how your buyers actually talk, every number that follows measures the wrong thing, and you are far better placed than we are to spot that.
What if the audit finds something urgent before delivery?
If we find something actively costing you visibility right now, a blocked crawler or a wholesale rendering failure for instance, we flag it immediately rather than waiting for the report. The report still lands at the scheduled time.
Can I pay after delivery instead of upfront?
No. The report is released once payment has been received. The order is: we agree the scope, we send the invoice, we do the work, and the finished PDF goes out once the invoice is paid. This is standard for fixed-scope diagnostic work. You can cancel any time before delivery, and if you have already paid we return the full amount.
Can the audit cover multiple sites?
The price covers one domain. If you operate multiple distinct sites, each needs its own audit and its own prompt set. If you operate one site with several country subfolders or subdomains, that's still one audit; we'll discuss scope during intake.
Ready to start the audit?
Send a short message describing your site and your competitors. We'll confirm the scope and start the intake within 24 hours.
Want the detail first? See what the audit covers.