How AI decides who to recommend
When someone asks ChatGPT which brand to buy, which service to use, or which company to trust, they expect a confident answer. That answer has to come from somewhere.
AI engines are not looking at a list of websites and picking the highest-ranked one. They are synthesising information from training data, real-time retrieval, and structured signals to form a judgment about which brand is most credible, most relevant, and most clearly identifiable in the context of the query.
The brands that get recommended are the brands that give AI engines the most reason to be confident. Confidence comes from clarity — clarity about what the brand is, what it offers, what it knows, and that others recognise it as an authority.
That confidence is built through six specific signals. Most brands have invested in none of them deliberately.
The six signals
Structured data and schema markup
Structured data is the machine-readable layer of your website. It uses standardised vocabulary — most commonly Schema.org — to tell AI engines exactly what your brand is, what your products or services are, where you operate, and what you are known for.
Without structured data, AI engines must infer this information from unstructured content. Inference introduces uncertainty. Uncertainty reduces the likelihood of a confident recommendation.
The highest-impact schema types for AEO are Organisation, Product, Service, and FAQ. A brand-facts.json file — a machine-readable summary of your brand identity hosted at your domain root — is an additional AEO-specific signal that few brands have implemented.
Brand entity clarity and consistency
In AI search, your brand is an entity — a discrete, identifiable object in the AI's understanding of the world. For AI engines to confidently recommend you, they need a clear, consistent picture of what that entity is.
That means your brand name, description, category, URL, and key attributes need to be described consistently across every platform where your brand has a presence — your website, social profiles, directory listings, press coverage, and third-party reviews.
Inconsistency is a confidence problem. If your brand is described differently across sources, AI engines cannot build a coherent entity model — and without a coherent entity model, they will default to a competitor they can describe with more certainty.
Citable, answer-first content
AI engines source their answers from content. But not all content is equally citable. Content that gets cited is content that clearly and directly answers a specific question — content structured to be extracted and referenced without requiring the reader to interpret or summarise it first.
This is answer-first content: content where the answer appears immediately at the top, before context, caveats, or background. Content with clear headings that signal what question each section addresses. Content with FAQ sections using proper schema markup.
Most brand content is written to engage human readers and rank in Google. That content is often too narrative, too promotional, and too vague for AI engines to confidently cite. Answer-first content is different in structure, tone, and intent — and it is the content that gets referenced.
Topical authority
A single page that answers one question does not make an authority. AI engines recognise topical authority through breadth and depth of coverage — a brand that has published comprehensive, structured answers across the full range of questions in its category.
This is the function of an Answer Hub: a structured library of citable content that covers every significant question a customer in your category might ask. The brands that have built this are treated by AI engines as the go-to reference for their topic. The brands that have not are treated as peripheral sources at best.
Topical authority is the signal that takes the most time to build — and is therefore the most valuable competitive moat once established.
Third-party mentions and citations
AI engines weight information from independent third parties more heavily than information sourced directly from a brand's own website. When journalists, reviewers, industry publications, and customers describe your brand consistently, that signals external validation.
This does not mean you need a traditional PR programme. It means your brand needs to be present, consistent, and positively described in sources outside your direct control: press coverage, review platforms, industry directories, and mentions in relevant community spaces.
The quality and relevance of third-party sources matters more than volume. One citation from a credible industry publication outweighs dozens of directory listings.
Technical content accessibility
AI engines cannot cite content they cannot read. Content locked behind login walls, rendered exclusively in JavaScript without server-side fallback, buried in PDFs, or blocked by robots.txt rules is invisible to AI crawlers.
Technical accessibility for AEO means ensuring your key content — especially your Answer Hub pages and product or service descriptions — is accessible to web crawlers, loads without JavaScript dependency, and is not inadvertently blocked by crawl rules.
This is the foundational signal. Without technical accessibility, none of the other five signals can be read. For most brands this is a maintenance issue rather than a major gap, but it should be the first thing checked in any AEO audit.
What most brands are missing
In practice, when we audit brands against these six signals, the pattern is consistent. Technical accessibility is usually adequate. Third-party mentions are often present but inconsistent. Everything else — structured data, entity clarity, citable content, and topical authority — is typically either absent or insufficiently developed to drive confident AI recommendations.
Strong content, weak signals. Most brands have written a lot. Very little of it is structured for AI citability. The content exists but the machine-readable layer does not. That is an AEO problem — and it is fixable.
The implication is that most brands are one to three months of focused work away from meaningfully improving their AI visibility. The signals are known. The methods are established. The gap between most brands and strong AI visibility is not capability — it is awareness and execution.
Audit your signals
Before building anything, establish your baseline. Run 20 purchase-intent prompts across ChatGPT, Perplexity, and Google AI and record whether your brand appears. That is your current visibility score — the number that everything else is designed to move.
Then audit each signal systematically:
The Found By AI audit runs this scoring automatically across 20 purchase-intent prompts and delivers a prioritised gap analysis with six content assets in 60 seconds. It is the fastest way to establish your baseline before beginning AEO work.
Frequently asked questions
AI engines recommend brands based on six key signals: structured data and schema markup, brand entity clarity and consistency, citable answer-first content, demonstrated topical authority, third-party mentions and citations, and technical content accessibility.
Structured data is the single highest-leverage signal. It makes your brand identity, products, and expertise machine-readable. Without it, AI engines must infer what your brand is from unstructured content, which reduces their confidence and therefore their willingness to recommend you.
Partially. Strong domain authority and quality backlinks signal credibility that AI engines consider. But SEO authority alone does not produce AI recommendations. The specific AEO signals — structured data, entity clarity, and answer-first content — must also be present.
Technical improvements like structured data can influence AI recommendations within weeks. Meaningful, sustained improvement in overall visibility score typically takes 60 to 90 days of consistent implementation across all six signals.
The six signals apply across ChatGPT, Perplexity, and Google AI Overviews, though the relative weighting differs. Structured data and entity clarity matter on all three. Topical authority is particularly important for Perplexity, which relies heavily on real-time retrieval. Google AI Overviews weight existing search authority more heavily than the other two.