The Citehelm AI Visibility Audit
The AI Visibility Audit shows whether ChatGPT, Perplexity, Google AI Overviews and Claude mention and cite your company when buyers ask about your category — in what context and tone, how often compared with your competitors, which sources the engines trust on your topics, and whether AI crawlers can read your site at all. You get the research, an optimisation strategy built on it, and a concrete implementation plan.
Six questions every marketing team should be able to answer now.
- Do AI assistants mention or cite us when buyers ask about our category — on which engines, for which questions?
- When they do, in what context and tone: recommended, merely listed, or compared unfavourably?
- How often are we named compared with our main competitors?
- Which sources do the engines trust most on our topics — and can we get into them?
- Can AI crawlers actually read our site, or is something blocking them?
- What exactly should we do, and in what order, to be mentioned more often and in a better light?
What we examine.
Can AI read you?
For each AI crawler — GPTBot, ClaudeBot, PerplexityBot and others — we check robots rules, CDN and hosting blocks, response codes, whether your content is readable without JavaScript, and your structured data. If crawlers cannot reach your pages, no content strategy will help, so this comes first.
Are you in the answer?
30–40 questions across the buyer journey, asked on four engines, three times each. We record whether you are cited (your page is a source) or only mentioned, the context and tone of every mention, and your share of voice against your competitors.
Who do engines trust?
For every question cluster we map the domains the answers are built from — directories, trade media, associations, review sites, competitor blogs — and separate the sources that are open to you from those that are locked.
What do your pages earn?
Which of your pages get cited and which never do, which buyer questions have no page at all, and whether engines recognise your brand correctly — name confusion, sibling domains, missing or inconsistent profiles.
Real questions, real engines, repeated.
We ask the engines the questions your buyers ask, in their language, on the consumer products they use — not on developer APIs, which can answer differently. Every question runs three times, because AI answers vary from run to run; a result counts when it holds in at least two of three. Your competitors are measured on exactly the same questions, so the comparison is fair. More on the methodology →
Research, strategy, and a plan you can act on.
Research report
Findings per pillar, your competitor comparison and the source map — about 10–15 pages.
Optimisation strategy
Where you can win, which sources to get into, which content to create or fix, which technical issues to remove.
Implementation plan
A prioritised action list with effort estimates, split into what Citehelm can do and what your team does.
Data and walkthrough
The raw measurement data and a 45-minute call to go through it together.
Not ready for an audit? Start with a Snapshot.
Send us your domain, your market and three questions a buyer would ask an AI assistant about your category. We add seven more, ask all ten on ChatGPT, Perplexity and Google AI Overviews, and check whether AI crawlers can reach your site. Within 48 hours you get a two-page PDF: where you appear, where you don't, and who appears instead of you.
A Snapshot is a single sample, not a measurement. It tells you whether there is a problem; the Audit tells you why, how big it is, and what to do about it.
| Snapshot | Audit | |
|---|---|---|
| Questions | 10 | 30–40, across the buyer journey |
| AI engines | ChatGPT, Perplexity, Google AI Overviews | The same three plus Claude |
| Repetitions per question | 1 (a sample) | 3 (a measurement) |
| Context and tone of mentions | — | Yes |
| Share of voice vs competitors | Who appears instead of you | Full comparison, per engine and topic |
| Source map | — | Yes |
| AI-crawler access | Quick check | Full technical review |
| Diagnosis and recommendations | — | Strategy and implementation plan |
| Delivery | 2-page PDF in 48 hours | Report + plan + call in ~10 working days |
| Price | Free | €500 launch price |
Do it yourself — or let us do it.
The implementation plan is written so your own team can carry it out. If you would rather we did, Optimisation runs in two phases: three months of implementation — the technical fixes, the content, the placements on trusted sources, with a measurement report every month — followed by monthly monitoring that keeps what you have won and catches what starts to slip.
Citation changes take weeks to show, which is why implementation hands over to monitoring instead of simply ending. If you start within 30 days of the Audit, its fee is credited in full. We commit to the work, not to citations.
Simple, published prices.
- 10 buyer questions
- ChatGPT, Perplexity, Google AI Overviews
- AI-crawler access check
- Who appears instead of you
- 30–40 questions × 4 engines × 3 runs
- Context, tone and share of voice vs competitors
- Source map and AI-crawler review
- Strategy and implementation plan
- 45-minute walkthrough
- Technical fixes and content
- Placements on the sources engines trust
- Monthly measurement report
- Where we can't access your site, we hand your team the exact changes
Prices exclude VAT.
About ten working days.
- Intake call (30 minutes) — your markets, languages, product lines, buyer questions and competitors.
- Question set — we draft the 30–40 questions and agree them with you within two days.
- Measurement — four engines, three runs per question, plus the crawler and source analysis.
- Analysis — diagnosis, strategy and the implementation plan.
- Walkthrough — 45 minutes through the findings and the plan.
What we will not promise.
AI answers are stochastic: the same question can produce different sources on different days. Nobody can guarantee a citation, and you should distrust anyone who does. What we do promise is a careful measurement, an honest diagnosis, and a plan whose effect we can measure again.
Direct questions, direct answers
- How is this different from an SEO audit?
- An SEO audit asks whether you rank in a list of links. This audit asks whether AI assistants name and cite you when they write an answer, and why. The two overlap on technical basics, but the measurement is different: we ask the engines real buyer questions and record what they say, which sources they use, and who they recommend.
- Why ask every question three times?
- Because AI answers change from run to run. A single answer is an anecdote; the same result in at least two of three runs is a finding. That is also the main difference between the free Snapshot and the Audit.
- Which AI engines do you cover?
- ChatGPT, Perplexity, Google AI Overviews and Claude, measured on the consumer products your buyers actually use rather than on developer APIs, which can answer differently.
- Do you need access to our systems?
- No. Everything can be measured from the outside. Read-only access to your server logs, Cloudflare or Google Search Console makes the crawler analysis more precise, but it is optional.
- Which languages and markets?
- We work in English, German, Hungarian and Portuguese, and measure each market in its own language — German buyers ask German questions and get answers built from German sources.
- What happens after the Audit?
- You can carry out the plan with your own team. If you would rather we do it, the Optimisation service implements it over three months and then moves to monthly monitoring. If you start within 30 days, the Audit fee is credited in full.
- Can you guarantee that AI engines will cite us?
- No, and nobody honestly can: generated answers are stochastic. We commit to the work and to transparent measurement of what it changes.
- How long does it take?
- About ten working days from the intake call: prompt set agreed in the first two days, measurement, analysis, then the walkthrough.
Request the Audit or a free Snapshot.
Three minutes. We reply within one working day; a Snapshot PDF arrives within 48 hours.