Anatomy of an AI citation audit: 30 queries, 5 agents, 150 retrievals.

The phrase “AI SEO audit” is becoming load-bearing, and like most fast-load-bearing phrases, it's rapidly losing meaning. Tools advertise “monitor your AI visibility” with dashboards that scrape ChatGPT for a handful of branded queries and call it a day. That isn't an audit. That's an inventory.

A real audit answers a strategic question: for the queries that drive bookings to my destination, am I in the answer set across the AI agents users actually use, and if not, what specifically do I need to change to get in? That requires more than a screenshot.

Why 30 queries, not 10

AI answers are non-deterministic. Run the same prompt through Claude or ChatGPT twice and you'll often get materially different recommendations. That's a feature, not a bug — the agents sample from a probability distribution over their retrieval set.

For a brand to claim it's “in the answer for X query,” you need a sample large enough to be statistically meaningful. Five runs is anecdotal. Ten is suggestive. Thirty queries across five agents — 150 independent retrievals — is the minimum we'll defend in writing. It's also small enough to be commercially viable as a fixed-fee audit.

The five agents — and why we test all of them

ChatGPT, Claude, Gemini, Google AI Overviews, Perplexity. These five between them account for the vast majority of AI-mediated travel research as of 2026. Each has a different retrieval bias:

• ChatGPT leans on a broad web index plus its training corpus; favours editorial and listicles.
• Claude leans heavily on long-form journalism — NYT, Atlantic, Economist, BBC, broadsheets.
• Gemini deeply integrated with Google's index; rewards strong domain authority and recency.
• AI Overviews the AI-summarised SERP itself; cites recently-crawled high-authority pages.
• Perplexity rewards primary sources, named trade press, and explicit citations.

A destination might win Perplexity (because we've placed it in Silver Travel Advisor and trade titles) but lose Claude (because the BBC and broadsheets haven't covered it). The audit tells you which agents you're winning, which you're losing, and what each one wants.

The query bank

We start with a standardised 20-query bank covering the planning patterns UK travellers actually use: “Where should we go in October for sun”, “Best family destinations under £200/night”, “Quiet alternatives to Tuscany”, and so on. You add 10 you specifically want to win — usually a mix of branded queries (“Things to do in Cyprus”) and category queries (“Best Mediterranean island for couples”).

You also nominate 3 named competitors. We run the full 30-query set against each of them too. The deliverable shows your citation rate, their citation rates, and which queries each competitor wins.

Citation gap → placement plan

The audit's payload isn't the rate. It's the gap analysis. For every query where you lose, we capture exactly which editorial pieces AI cited instead. That breaks down into three lists:

1. Publications we already write for — straightforward fix, plan the next pitch into them. Goes onto the retainer plan immediately.
2. Publications we have relationships with but haven't commissioned recently — a 6-to-12-week pitch window, usually pricing into the retainer.
3. Publications neither side has worked with — flagged for either organic-pitching or, if commercial logic supports it, a sponsored editorial conversation.

Plus a parallel list of feature angles that worked for competitors — the actual story shapes you could legitimately commission for your destination. Sometimes the gap is that we don't have a story under your name yet about a specific angle; the audit names that and we go and write it.

What the audit does NOT do

Audits don't move citation rates. Editorial placement does. The audit is a diagnostic — what to fix, in what order, with what story angles, in what publications. The retainer is what actually fixes it.

That's why the audit fee is credited against the first three months of retainer for clients who continue. We don't want clients buying audits and walking away with a fix-list they can't execute. We want clients who are buying a citation-rate improvement, and the audit is just the first step of that.

The full method page lays out the 5-day timeline and what each tier delivers in detail.