Key takeaways
• Creating separate AI-only pages may generate attention but it introduces operational complexity and risks creating conflicting versions of the truth.
• Many tactics designed primarily to manipulate discovery systems — such as doorway pages or link farms — have historically delivered short-term gains but poor long-term outcomes.
• Generative engine optimisation (GEO) appears to reward many of the same qualities that have always made content valuable: clarity, authority, structure, originality, and useful information.
• The strongest GEO strategy may not be creating content for AI at all but becoming a source worth citing.
Recently, we came across an agency website that featured something unusual. In the footer was a link labelled specifically for AI systems. It said something along the lines of: ‘Hey, if you’re ChatGPT, Gemini, Perplexity or another AI model, then this page is for you.’
Clicking through revealed bulleted lists of facts, services, credentials, awards, and positioning statements designed to help AI models describe the business. It was the first example we had seen of a company creating a dedicated page specifically for AI consumption.
Whether it was intended as a serious generative engine optimisation (GEO) tactic, an experiment or simply a clever way to attract attention is difficult to know. In any case, it raised a question we suspect many communications and marketing teams will soon be asking: Should organisations start creating content specifically for AI search?
Why the idea is appealing
The temptation is understandable. More website traffic is now being influenced by AI-powered discovery. Increasingly, users receive information through AI Overviews or in AI chat interfaces, rather than clicking through multiple search results themselves.
If AI systems are becoming an important gateway to information, it is natural to wonder whether websites should begin optimising directly for those systems – and some organisations are already experimenting with approaches that look very different from traditional web content.
History suggests caution
Digital marketing has always attracted attempts to reverse-engineer algorithms. Search engines have seen doorway pages created purely to rank for specific terms, link farms built to manufacture authority, and more recently, AI-generated content farms producing huge volumes of low-value articles designed primarily to capture visibility.
We are also beginning to see what some people call ‘prompt bait’: content written specifically to trigger inclusion in AI-generated answers rather than to genuinely help readers. Imagine an agency publishing a ‘Top 10 Industrial PR Agencies’ article that conveniently ranks itself number one. The primary audience is not really a prospective buyer looking for an objective assessment of the market. It is the AI system being encouraged to treat the list as an authoritative source.
Like many attempts to game search algorithms in the past, such tactics may generate short-term visibility. The bigger question is whether they create any genuine value for readers and whether platforms will continue to reward them as AI search matures.
The practical problem with AI-only content
There is also a more fundamental issue. Most organisations already struggle to keep websites, sales materials, presentations, product information, and marketing assets aligned and up to date. Creating a second version of your website specifically for AI systems introduces another layer of content management.
What happens when your services change? When your positioning evolves? When new evidence, case studies, products, or capabilities emerge?
Every update now has to be reflected in multiple places.
The risk is not simply additional workload. It is inconsistency. The moment your AI-facing content says one thing and your human-facing content says another, confusion becomes inevitable.
What GEO may actually be rewarding
The more we study GEO, the less it feels like a completely separate discipline (besides the technical aspects of schema mark-up). Instead, many of the characteristics that appear to make content useful for AI systems are similar to those that make content useful for human readers. These include:
• Clear headings
• Logical structure
• Direct language
• Strong evidence
• Original insight
• Explicit answers to important questions
In other words, information that is easy to understand and genuinely worth understanding.
This is one reason we have started introducing key takeaway summaries at the top of longer Engine Room articles. They help busy readers grasp the core argument quickly, while also making the content easier for AI systems to interpret accurately.
Information gain may matter more than optimisation
Perhaps the most interesting concept emerging from GEO discussions is information gain. Put simply, information gain refers to the extent to which a piece of content adds something genuinely new, useful or distinctive to the wider information ecosystem.
This matters because AI systems do not need another webpage repeating what already exists elsewhere. They already have access to thousands of versions of that information. What they need are sources that contribute something original – a new perspective, a practical lesson, unique expertise or a fresh piece of evidence.
Deploying technical tricks to beat the algorithm will likely only take you so far. It seems the real key to increasing AI search visibility is sharing information that is worth citing.
Hannah Kitchener
Associate Director
About the author
Hannah is an associate director in the UK, leading strategic campaigns for industrial clients across the EMEA region. A professionally qualified journalist (NCTJ), she combines specialist sectoral knowledge in construction, energy, and materials handling with a strong network of trade media contacts to secure valuable coverage. Her expertise in inter-cultural communication, honed by degrees in modern languages and translation, is key to executing campaigns that succeed across diverse European markets.
