To audit a brand’s visibility across AI systems, you need to test whether the brand appears for the questions potential customers actually ask—and record enough context to make the result reproducible. A single search is an anecdote. A controlled set of prompts, engines, dates and sources is a benchmark.
The method below is designed for organisations that already understand the commercial problem. It is not a promise that every person will receive the same answer. AI outputs can vary by system, time, location, account context and wording.
Keep participation, access and visibility separate: brand-visibility testing observes answers and citations. Technical-access testing examines whether relevant crawlers and retrieval systems can reach the website. One cannot substitute for the other.
Why the UK requires clear control boundaries
UK publishers and brands increasingly face separate decisions about conventional search inclusion, access by AI-related crawlers and appearance inside generated answers. A visibility audit should record these as different layers. A permission choice is not evidence of citation, and an absent mention does not prove a technical block.
For organisations operating across the UK and Europe, preserve the tested country, language, domain and route. Consent systems, regional redirects and edge security can change what is actually delivered even when one global policy file is published.
What should the audit measure?
| Dimension | Question | Evidence |
|---|---|---|
| Presence | Is the brand named at all? | Exact answer text and timestamp. |
| Prominence | Where and how strongly is it presented? | Order, wording and recommendation context. |
| Accuracy | Are the brand, offer and facts represented correctly? | Claim-by-claim comparison with authoritative sources. |
| Citations | Which sources support the answer? | Visible links, cited domains and landing pages. |
| Competitor set | Which alternatives repeatedly appear? | Named competitors by prompt and system. |
| Technical eligibility | Can retrieval systems access the relevant pages? | Separate robots policy and observed-access evidence. |
A six-step brand visibility audit
How to design the prompt set
The strongest prompts reflect real purchasing decisions rather than vanity checks. Start with five groups:
- Category: “Which providers offer [service] in the United States?”
- Problem: “How should a company solve [specific problem]?”
- Comparison: “Compare providers for [defined use case].”
- Trust: “Which [category] providers have evidence of [required capability]?”
- Branded accuracy: “What does [brand] provide, and who is it for?”
For local services, specify a genuine service area. For B2B and enterprise products, include the buyer role, operating environment, risk or integration need. Avoid prompts so broad that the answer set has no relationship to a realistic purchase.
Market: United Kingdom
System and mode: ______
Date/time: ______
Prompt ID and exact wording: ______
Brand mentioned: Yes / No
Prominence and context: ______
Competitors named: ______
Cited domains and URLs: ______
Factual errors or unsupported claims: ______
Screenshot/evidence reference: ______
Why technical access is tested separately
If a brand is absent, the cause may be insufficient authority, ambiguous information, weak relevance, answer volatility—or a technical access barrier. You cannot diagnose the cause from the absence alone.
Technical testing asks a narrower set of questions: what does robots.txt declare for distinct crawler identities, what site guidance is available, and what response does an honestly identified audit crawler receive? CDN and WAF behaviour can contradict the published policy. Conversely, a technically open site may still lack the evidence required to be selected.
This is why Digital Dominator reports declared policy and observed access separately, and why the paid technical audit does not claim to be a prompt-monitoring or competitor-ranking product.
Common mistakes that invalidate the result
- Testing one prompt once and presenting it as market visibility.
- Changing prompt wording between brands or reporting periods.
- Mixing US and overseas markets without recording the difference.
- Counting a navigational response as an unprompted recommendation.
- Treating every citation as positive without checking accuracy and relevance.
- Assuming robots.txt permission proves that infrastructure access succeeded.
- Assuming no mention proves that the website was blocked.
- Publishing a single score without preserving the underlying evidence.
When to commission a technical audit
A technical access audit is appropriate when the organisation needs to verify its crawler policy, diagnose CDN or firewall behaviour, investigate contradictory test results, or give a developer an exact remediation list. It is also useful before investing heavily in broader AI visibility work: there is little value building more content while an unintended access rule prevents relevant retrieval.
Verify the access layer
Digital Dominator’s UK audits provide human-reviewed policy and access evidence with developer-ready remediation. USD pricing is available at checkout.