What does an SEO and AI visibility audit actually include?
5 minute readBy Dan, Foundlab
An SEO and AI visibility audit should explain what was checked, what the evidence shows and what to do next. You should be able to hand its recommendations to someone and know how the result will be verified.
That sounds straightforward until a report arrives with a long list of warnings and no indication of which ones matter. A missing description on an old page and a blocked main service page should not receive the same attention. I look for the connection between the finding, the customer’s task and a practical change.
What a useful finding contains
- Observe
- The affected page and the evidence collected.
- Explain
- Why it matters, and what the evidence cannot tell us.
- Act
- A specific change, an owner and a priority.
- Verify
- The check that will show whether the change worked.
Start with the business and the demand
Before checking tags, establish what the business wants to be found for. Which services matter? Who buys them? Where can those customers be served? What do they ask before making contact? This keeps the audit connected to the work the business actually wants.
Search Console can show the queries and pages already generating impressions and clicks. Google Trends can help compare relative interest and changes over time. These sources answer different questions. A Trends index is not a monthly search-volume figure, and a new site’s limited Search Console history does not prove that demand is absent.
Foundlab’s demand map uses a shared anchor when comparing Trends terms. The report should still retain the geography, time period and method, and flag where low data makes a conclusion uncertain.
Check whether the pages can be found and understood
The technical review covers page responses, crawl permissions, indexing evidence, canonical addresses, internal links and the content a crawler can read. The on-page review asks whether titles, headings and copy accurately explain the offer. Mobile usability and page experience belong in the review too.
Indexing needs careful wording. A successful live fetch tells us a page can be retrieved from that test environment. Google’s URL Inspection report provides separate information about its indexed version. Neither observation guarantees a ranking for a particular query.
Where owner access is unavailable, the audit should say so. Public checks are useful, but they do not reveal private Search Console performance or a complete history of enquiries.
Record actual AI answers
An AI visibility review should include customer questions, the platforms tested, the date and the sources cited in each answer. Record whether web search was enabled and enough context to make the check repeatable. A brand-name question and a buying question test different things.
A business may be mentioned without being recommended. Its article may be cited in an answer that recommends someone else. Those are distinct outcomes. A single run is an observation, not a stable share-of-market score.
Technical access checks support this work. For example, OpenAI documents OAI-SearchBot as its search crawler and GPTBot as a crawler for potential training use. Treating every AI-related crawler as interchangeable can produce the wrong recommendation.
See what a finding looks like
This example comes from Foundlab’s own website before the author update. It shows the level of detail a reader should expect: the page, the observation, the reason for the change and the verification step.
Priority: clarity and trust
Show who writes the advice.
The article page displayed a publication date and reading time, but no named author. Its article markup identified the business as the author.
View the evidence and recommended change
- Page checked
- /insights/your-brand-name-might-belong-to-someone-else
- Evidence
- Public HTML inspected on 9 September 2026, before this update. The visible byline was absent; the BlogPosting author referred to the organisation.
- Why it matters
- A reader should be able to see who is responsible for first-person advice and follow a link to their relevant experience. This finding does not establish a ranking penalty.
- Recommended action
- Add a named byline, an About page and consistent author information in the article markup. Owner: site editor/developer.
- Verification
- Read the updated article, follow its author link, and check that its visible author and structured data agree.
An excerpt illustrating the report format, using a real observation on our own website. It is not a complete audit or a claim of improved rankings.
Leave with a usable order of work
A recommendation should name the affected page, the action and the person who can carry it out. Priority should reflect likely value, confidence and effort. Confirmed blockers belong ahead of speculative experiments. Explain what evidence would change that order.
For content, a topic list alone is not a finished deliverable. A useful brief explains the reader’s question, why the topic matters, the original information needed and the service page it supports. A publish-ready article should also have a title, description, relevant sources and internal links.
What Foundlab includes
The Findability Audit combines a demand map, technical and on-page checks, recorded AI answer checks, a prioritised action list, five articles ready to publish and a 30-minute walkthrough. The published price is A$1,500–A$2,500, with scope and price confirmed before commitment. Delivery is 7 business days from the agreed start.
Implementation is scoped separately. Rankings, AI recommendations and revenue are not guaranteed deliverables. The first call establishes whether an audit is a sensible use of your money.
You can begin with the free visibility checklist, read about the person doing the work, or inspect the published case studies before deciding.