What BrightonSEO taught me about building visibility in AI search

I recently attended my first BrightonSEO, with sessions exploring how AI is changing search, content and the way people discover brands. There were perspectives from speakers including Laura Smith, Gijs de Groot, Ramona Joița, Andrew Chadwick and Josh Blyskal, alongside practical examples of how brands are researching and improving their AI visibility.

My biggest takeaway was that AI search is pushing us to understand customers in more detail. Keywords and rankings still matter, but we also need to consider the situations people are in, the questions they ask and the evidence that helps a brand become a relevant recommendation.

Customer context needs a bigger role in search strategy

Customer context was one of the strongest themes across the sessions. A brand might be well known within a category but rarely appear when someone asks for a recommendation for a particular situation.

Laura Smith illustrated this through a small study involving Free Soul. The brand appeared in all five tests for “best protein brands for women”, but only one of 15 tests relating to strength or active use. These were observations from a limited study, but they highlighted a useful distinction: visibility for a broad category does not automatically mean visibility for the situations in which customers might choose your product.

Adding detail about a customer’s needs can change the brands an AI platform recommends. “Running shoes” is a broad request, while “a cushioned daily trainer with a wide toe box for road running rather than trails” gives a much clearer picture of what someone needs.

For marketers, that means looking beyond broad category terms. Search data, reviews, customer service questions and CRM insights can help identify the circumstances that influence a decision. Our content then needs to explain clearly how the product or service fits those circumstances.

Prompt tracking is only useful if we ask relevant questions

Several talks explored prompt tracking: regularly testing questions in AI platforms to understand which brands appear, what the answers say and which sources they cite.

Gijs de Groot challenged how much confidence we should place in AI visibility scores when the underlying prompts do not reflect the audience. A score might look positive while missing the questions that matter most to potential customers.

A useful starting point is to track three types of questions:

• Branded questions: Does AI describe the brand, its services and its features accurately?
• Competitor comparisons: Does the brand appear when people compare options or ask for alternatives?
• Non-branded discovery questions: Is the brand recommended when someone describes a need without naming a provider?

These questions should come from real customer needs and the search data we already have, including Search Console and Bing Webmaster Tools. AI can help refine the wording or suggest variations, but those suggestions still need to be checked against what customers actually ask.

The value comes from what we learn. If a client repeatedly appears for broad questions but is absent for a priority use case, that gives us a specific gap to investigate.

AI visibility extends beyond our own websites

Earned media and external reputation also featured throughout the talks. AI answers can bring together information from brand websites, publications, reviews and communities, making the wider online picture relevant to how a brand is presented.

For agencies, this creates a stronger connection between SEO and PR. Reviewing the publications, comparison pages and communities cited for priority questions can help identify where credible external coverage could support the client.

It also makes consistency important. If product details, service descriptions or brand positioning differ across sources, customers may receive a confusing picture.

The practical opportunity is to work across teams: improve the information on the website, identify relevant external sources and ensure the brand’s key facts are clear wherever people encounter them.

Product pages need to explain who a product is for

The ecommerce sessions highlighted the importance of product detail pages. These pages need to help someone understand whether a product suits their needs, as well as provide specifications and a route to purchase.

The SEED framework offered a useful way to approach this:

• Situational relevance: Who is the product suitable for, and where might it be less suitable?
• Example-led content: How is it used in everyday life?
• Enriched product data: Are specifications, price and availability clear and consistent with product feeds?
• Distinct brand voice: Does the content communicate the brand’s perspective and expertise?

For ecommerce brands, reviews and support questions can reveal what people need to know before buying. Those insights can strengthen product pages while supporting the buying guides and category content around them.

Technical accessibility still matters

The discussions around AI search also reinforced the importance of technical SEO. Useful information needs to be accessible before it can support discovery.

An important check is whether key content appears in the page’s HTML or depends entirely on JavaScript. Product reviews, size guides, specifications and supporting links may be visible to a customer in the browser but harder for some crawlers to access.

For priority pages, reviewing the HTML and testing with JavaScript disabled can help identify potential gaps. Clear answers, accessible supporting detail and relevant internal links remain practical areas to focus on.

Creating content with AI is not necessarily faster

One of the content lessons that stayed with me was that using AI does not automatically make good content quicker to produce. A one-line prompt gives the tool very little understanding of the audience, the argument or the experience behind the piece.

Quality still takes a clear brief, research, refinement and editorial judgement. Working in stages, beginning with an outline and developing sections individually, gives us more opportunity to shape the result.

The talk’s PEEL framework brought this back to four useful ingredients:

• Pain: Address a specific reader problem.
• Evidence of lived experience: Include first-hand insights, examples and stories.
• External validation: Support the argument with credible research.
• Links: Connect readers to useful sources and relevant internal content.

AI can help organise and develop material, but the distinctive part comes from what we contribute. Real experience, customer insight and an informed point of view give readers a reason to choose our content.

Measurement needs to consider the wider customer journey

The ecommerce discussions also raised a useful measurement question: what happens if someone discovers a brand through AI, then returns through Google or visits the website directly?

That journey may not appear as an AI referral in analytics. Looking only at direct AI traffic can therefore leave part of the influence unseen.

It does not mean every increase in branded search or direct traffic should be attributed to AI. It means we need to assess several signals together, including tracked visibility, citations, referral traffic, branded demand and commercial results.

Conclusion

I came away from BrightonSEO with several practical ways to develop our client strategies: build prompt studies around genuine customer needs, inspect the sources behind recommendations, strengthen product and service information, and connect website improvements with external credibility.

The sessions also reinforced the value of work we already do in SEO. Understanding intent, making content accessible and helping people make informed decisions remain essential.

For me, the next step is to bring more customer context into that work. When we understand the situations in which people choose a brand, we can create clearer content, ask better research questions and make more useful recommendations to clients.