AI Visibility 2026
What companies need to know now.
3 October 2026 · 5 min read
Start with the decision your customer needs to make
A useful visibility analysis starts with the questions people ask before choosing a provider. Are they looking for a local business, comparing specialist services or checking a particular brand? Treat these situations separately. Collect questions from customer conversations and your sales team, add the market context and check that the wording is neutral. An answer to a question that names your brand means something different from a recommendation where your company was not specified. First decide which customer decision your measurement should represent.
Separate mentions, recommendations and sources
A brand can appear in an answer without being recommended. A linked page can serve as a source while the answer names another provider. Read the full answer and record the context: who appears, why, and with which sources? Observe ChatGPT, Claude, Gemini and Perplexity separately. An aggregate can help you navigate the results, but it cannot replace the individual question and system. That specific observation provides the basis for a decision.
Establish a comparable baseline
Document the questions, market, language, systems and measurement date. Repeat measurement using the same reviewed question set before interpreting a change as a trend. Missing or failed answers are missing evidence and must not appear as zero visibility. Check that both measurements are complete and comparable. If a question changed, interpret that difference separately. A sound baseline helps prevent action based on a chance observation.
Check the information behind the answer
Review the cited pages and the statements about your company. Are your offer, location, specialisation and contact details clear and consistent? Can readers verify your factual claims? Does your content answer a specific customer question clearly, with reliable evidence? Use the observed sources as a starting point for this review. A content change offers no guarantee of an AI mention. Prioritise a specific information gap instead of rebuilding content around a supposed GEO formula.
Turn the finding into an action you can check
A useful recommendation identifies the question, observed answer, relevant source and proposed change. Agree the target page, accountable person and approval. Then document what was actually implemented and when. Your company, agency or explicitly appointed partner carries out the work. A later comparable measurement shows an observed change; on its own, it does not prove the cause. Another measurement helps you check whether the signal persists.
Use search data as supporting context
Search Console can show which questions and pages matter in organic search. Aggregate GA4 data can add context about the use of organic landing pages. These signals do not replace observations of AI answers or automatically establish where an individual enquiry came from. Start with a clear brief: a reviewed question set, separate systems, a documented baseline and one prioritised information gap. Make decisions from the available evidence and state openly what has not yet been established.