Most brands still have no idea what ChatGPT, Perplexity, or Gemini say about them when a potential customer asks a relevant question. Not a rough idea - no idea at all. They have not checked. They have not asked anyone to check. And in the meantime, those AI tools are generating answers about their products, their positioning, and sometimes their competitors, thousands of times a day.
Yoast recently drew attention to this gap with the launch of an AI brand insights scan, giving eligible users a report on how their brand appears across ChatGPT, Perplexity, and Gemini. The tool itself is theirs. The problem it points to belongs to every brand with a web presence.
The Passive Presence Problem
Traditional SEO gave brands a degree of control. You could see your rankings, check your visibility scores, and track movement over time. AI-generated answers do not work like that. There is no rank tracker for what Gemini says about your brand when someone asks which accounting software is best for a UK small business. You either know what it says, or you do not.
The issue is not that AI tools are malicious. It is that they are drawing on whatever they have been trained on or can retrieve, and that content may be outdated, incomplete, or written by a competitor. If your brand has weak structured content, sparse third-party coverage, or inconsistent messaging across channels, that is what gets reflected back.
Being absent from AI-generated answers is one problem. Being present but described inaccurately is arguably worse. A customer who asks Perplexity about your pricing, your features, or your suitability for their use case is going to act on what it tells them. If that information is wrong, you lose the sale without ever knowing the conversation happened.
Why a Baseline Matters Before Anything Else
There is a temptation in GEO and AEO discussions to jump straight to tactics - structured data, topical authority, entity building, authoritative citation sourcing. All of that is valid. But it assumes you know where you are starting from. Brands that skip the audit phase end up optimising in the dark.
A proper AI visibility baseline should answer at minimum three questions. First, does your brand appear at all when relevant questions are asked across the major AI platforms? Second, when it does appear, is the description accurate and favourable? Third, who else is being cited instead, and why might the model be preferring them?
These are not questions you can answer with a single prompt. They require systematic querying across different intent types - informational, comparative, transactional - and across different AI platforms, because the answers vary. ChatGPT and Perplexity use different retrieval mechanisms and training data. Google AI Overviews are tied more directly to the web index. Gemini sits somewhere between the two. Your visibility profile is likely different on each.
What AI Tools Are Actually Evaluating
When an AI model generates a response that includes a brand recommendation or description, it is not doing a live SEO crawl. It is drawing on a mixture of training data and, where retrieval-augmented generation is in play, current web content. The signals that influence whether your brand gets included are different from traditional ranking factors.
Consistency matters more than most brands realise. If your website describes your product one way, your PR coverage describes it another way, and your review profiles describe it a third way, the model has conflicting signals and may either omit you or produce a muddled description. Entities that appear frequently, consistently, and in credible sources tend to win the citation.
This is where the GEO discipline intersects with brand communications more broadly. It is not purely a technical SEO problem. The tone of your press coverage, the specificity of your case studies, the quality of third-party reviews - all of it feeds into how confidently an AI model can describe what you do and who you serve.
Monitoring Is Not a One-Off Exercise
Running an initial brand scan is useful. Treating it as a one-time activity is not. AI models are updated. New content enters the retrieval pool. Competitors publish content designed to position them more favourably. A brand that was well represented in AI answers three months ago may not be today.
Regular monitoring should become part of the same rhythm as checking organic search performance or reviewing paid media metrics. The cadence can be monthly for most brands, more frequent for those in competitive categories or in the middle of a brand refresh or product launch. The key is to track changes, not just snapshots.
It is also worth monitoring across query types, not just branded queries. Knowing how you appear when someone asks a non-branded question that you should be answering - what CRM is best for a construction company, for instance, or which UK payroll provider suits a 50-person team - is often more commercially significant than tracking branded mentions alone.
Turning Visibility Data Into Action
Once you have a clear picture of your current AI visibility, the remediation work becomes more focused. If you are absent from answers entirely, the priority is typically building topical authority through well-structured, comprehensive content that clearly defines what you do, for whom, and why. If you are present but described inaccurately, the focus shifts to correcting inconsistencies across owned and earned channels.
If competitors are being cited ahead of you, it is worth analysing exactly what content is being referenced in those citations. Often, it is a specific type of asset - a comparison guide, an FAQ page, an independently reviewed piece of coverage - that you do not currently have or have not made easy for AI systems to find and attribute.
The broader point is that AI visibility is now a measurable, manageable channel - not just an abstract concern. Tools exist to assess it, frameworks exist to improve it, and the brands that treat it systematically will be better positioned as AI-assisted search becomes the default way people find and evaluate products and services. The ones waiting to see how it plays out are already behind.