← Field Notes · October 13, 2026

A practical AI search visibility audit you can run without buying another dashboard

Published October 13, 2026
Key Takeaways
A repeatable way to check whether your business appears in AI-assisted research, diagnose why it is missing, and decide what to fix first.

Before you pay for an AI visibility platform, spend an hour finding out what you actually need to measure. A dashboard can make a weak measurement process look sophisticated. It cannot tell you whether the questions you are tracking reflect how a customer buys.

Start with the buying decision, not a list of prompts

Write down five situations that make somebody look for a business like yours. A homeowner might be comparing roof replacement against repairs, looking for a contractor who works with insurance, or trying to understand whether a quote is reasonable. A software buyer might be checking integrations, implementation time, data migration, or the difference between two vendors. These are different decisions, so treating all their prompts as one visibility score obscures the useful information.

For each situation, create two or three realistic questions. Avoid inserting your brand name into the question. If your brand is already there, you are testing whether the system can repeat your name, rather than whether it recommends you when a buyer does not know you exist.

Build a small, defensible baseline

Use a spreadsheet with the following columns: date, search product, exact question, geographic context, whether the business appeared, whether its own website was cited, which competitors appeared, cited source URLs, and the answer’s main claims. Record the search product and date because AI answers change. Do not average one ChatGPT answer, one Google AI Overview, and one Perplexity answer into a single number without labeling the different systems.

Run the same question more than once across separate sessions if possible. A single appearance is an observation, not proof that your brand consistently wins that query. Keep the original answer or screenshot for later review. When a product personalizes results or uses location, document that limitation instead of pretending the test is perfectly controlled.

Diagnose the missing evidence

Suppose your company never appears for “roof replacement financing options near me.” Look at the sources the search product actually used. Are competitors explaining payment options with real terms? Are they mentioned on independent local websites? Does your own site have a useful financing page, or only a sentence buried in a footer? Is the page accessible to crawlers, linked from relevant service pages, and clear about geographic coverage?

The fix might be a better service page, a stronger explanation of financing eligibility, clearer company facts, or legitimate third-party coverage. It is unlikely to be adding the phrase “best AI-recommended roofing company” to every paragraph.

Prioritize work with a decision table

Finding What to check Reasonable next step
Business absent, competitors cited Competitor source pages and missing buyer information Publish a genuinely useful answer on the appropriate existing page
Business mentioned, facts wrong Website, listings, profiles and outdated third-party references Correct first-party facts and request corrections where possible
Website not cited Indexability, internal links, original evidence and source relevance Fix technical access and improve the page’s actual usefulness
Mentions without enquiries Query intent, referral data and conversion path Measure commercial outcomes instead of chasing mentions alone

Repeat the test without manufacturing a success story

Keep a stable core set of questions and review it monthly. You can add new questions when products, markets or customer concerns change, but do not quietly replace the original set every time performance looks disappointing. Record changes made to the website alongside the observations. Correlation is useful for prioritization; it is not proof that one paragraph caused a citation.

Use Google Search Console to track search impressions and clicks, your analytics platform to assess enquiries, and the AI-answer log to understand where your company is represented. Each tool answers a different question. The practical goal is to make your business easier to find, evaluate and contact, not to produce an impressive-looking AI visibility percentage.

For the underlying search principles, see Google’s helpful content guidance and its guidance on AI search. If you need help prioritizing the work, compare NetPageTwo’s search programs.

Like this:

Discover more from Net Page Two

Subscribe now to keep reading and get access to the full archive.

Continue reading