What is included in an AI-visibility audit?

Prompt testing across the AI tools your customers actually use

The foundation of an AI-visibility audit is running the real questions your buyers ask through the AI tools they use, then recording what comes back. This is the part most "audits" skip — and without it, everything else is guesswork.

We start by building a prompt set from your real demand: category questions ("best commercial roofer in Calgary"), comparison questions ("is X or Y better for small business payroll"), and problem-led questions ("who can fix a leaking flat roof fast"). A good set is 30 to 100 prompts, weighted toward how people actually phrase things in chat, which is longer and more conversational than a Google search.

Each prompt gets run through several engines, because they don't agree with each other. ChatGPT, Perplexity, Google's AI Overviews, Gemini, and Microsoft Copilot pull from different indexes and weight sources differently — you can be invisible in one and recommended in another. Running a single tool gives a misleading picture.

For every prompt we log three things. First, presence: are you named at all? Second, framing: what does the AI actually say about you — is it accurate, flattering, outdated, or wrong? Third, the citations: which web pages did the model link to or draw from when it built that answer. That third point is the lever, because it tells you which sources AI trusts on your topic.

Because AI answers vary between sessions, the audit runs each prompt more than once and captures screenshots. The deliverable is a scored matrix — prompt by tool — showing exactly where you appear, where a competitor takes the spot, and where the AI gives no clear recommendation at all (your easiest openings).

Competitor benchmarking and citation-source analysis

The second piece answers the question that actually changes your strategy: when the AI doesn't recommend you, who does it recommend, and where did that answer come from? An AI-visibility audit isn't useful in isolation — it's useful relative to whoever is winning the answer today.

For each prompt where a competitor appears, the audit records which competitors recur and the framing the AI uses for them ("known for fast turnaround," "affordable option," "trusted local provider"). Patterns emerge fast: often two or three names dominate a category not because they're better, but because their content is structured in a way AI can lift cleanly.

Then we trace the citations. We pull the specific URLs the models cite and categorize them: the competitor's own pages, directories and listings, review platforms, industry publications, Reddit and forum threads, and Wikipedia-style references. This shows you the "source map" for your category — the handful of places AI consistently trusts. If Perplexity keeps citing a roundup article or a particular directory, getting represented there matters more than another blog post on your own site.

We also check what's being said about you off your own website, because AI weighs third-party corroboration heavily. Inconsistent business names, an outdated address, thin or missing review presence, or no mentions in any industry source all suppress how confidently AI will recommend you.

The output is a gap analysis: for each high-value prompt, here's who wins, here's the exact source feeding that answer, and here's the realistic path to displace or join them. That turns a vague "we want to show up in ChatGPT" into a concrete list of pages to create, sources to earn placement in, and facts to correct.

Technical and content readiness AI crawlers depend on

The third part checks whether AI systems can actually find, read, and trust your content — because even strong pages get ignored if machines can't parse them cleanly. This is the on-site, fixable layer of the audit.

On the technical side, we verify your pages are crawlable by AI bots (GPTBot, PerplexityBot, Google-Extended and others aren't always allowed by default in robots.txt), check for an llms.txt file, and confirm your content renders as real HTML rather than being locked behind JavaScript the crawlers may not execute. We review schema markup — Organization, LocalBusiness, Product, FAQ, and Article — because structured data hands AI clean, unambiguous facts about your name, location, services, and pricing instead of making it guess.

On the content side, we assess how "extractable" your pages are. AI favors content that answers a question directly in the first sentence, uses clear headings phrased as real questions, states facts plainly, and shows recency. Long, vague, keyword-stuffed pages that bury the answer perform poorly because the model can't lift a confident statement from them. We flag which of your key pages read well for AI and which need restructuring.

We also check the trust signals AI leans on: a real About page, named people, credentials, consistent NAP (name, address, phone) across the web, and visible, recent reviews. AI is conservative about recommending businesses it can't corroborate.

The section ends as a prioritized fix list, ranked by effort versus impact — for example: open robots.txt to AI crawlers (quick, high impact), add LocalBusiness schema, rewrite three service pages to lead with the answer, and earn a citation in the directory that keeps appearing in your competitors' answers. That's the bridge from audit to actual visibility gains.

What you receive, and how it fits into AI search work

You should walk away from an AI-visibility audit with a clear baseline and a do-this-next plan — not a number with no context. If an "audit" is just a single visibility score, it can't tell you why you're invisible or what to do about it.

A complete deliverable includes: the prompt-by-tool scoring matrix with screenshots, the competitor and citation-source gap analysis, the technical and content readiness checklist with pass/fail per item, and a prioritized roadmap that ties each recommendation to specific prompts and pages. Good audits also set the baseline you'll track against, since AI visibility only means anything when you measure movement over time.

On pricing, treat the audit as the entry point to AI search (GEO) work, which in the Canadian market typically runs as part of a retainer rather than a one-off. AI search optimization usually sits within a single-channel scope starting around $1,500/month, or inside a fuller program where SEO commonly runs $2,500 to $7,500/month. Many businesses fold the audit into a broader SEO and content engagement because the fixes — schema, page rewrites, earning citations, reviews — overlap heavily with traditional SEO. The audit just reframes that work around what AI specifically rewards.

This is exactly the kind of project where one team handling SEO, content, web development, and AI search together pays off, because almost every audit fix touches more than one discipline: a content rewrite, a schema deployment, a robots.txt change, a review push. SearchPod runs the audit, owns the fixes across those channels, and reports the AI-visibility movement transparently — with you keeping ownership of the accounts and data throughout. If you want to see where you currently stand in AI answers, an audit is the right first step.

Related questions