5 ways to optimise for AI search

The fundamentals are simple. To perform well in LLMs like ChatGPT, Claude or Gemini, your content must provide unique, up-to-date information that is accurate, easy to understand, and presented in a way that is visible to crawlers.

As Dena Warren highlighted: "A lot of e-commerce brands in the past have just used manufacturer copy, and you have the same product copy across 30 e-commerce websites... nowadays you really need to have original product and category content where possible."

Warren recommends enriching product descriptions with structured specifications, specific dimensions, compatibility details, and clear use cases.

Implementing clear page structures using features like subheadings, author bios, and tables makes your information easily digestible for both real-world users and AI models. The additional use of structured data in the form of schema is less clear-cut in terms of AI impact, but it remains a useful consideration. 

“Schema implementation isn’t quite the magic bullet for LLM and AI optimisation that some claim. But it remains a critical component of feeding the web’s understanding of your company, website, and pages, and how they relate to the wider web of entities, topics and things.” Says Candour’s own Principal SEO Specialist, Brendan Bennett. 

“Particularly for e-commerce, and as agentic capabilities increase, the role of structured data is likely to grow much more than it will be diminished.”

To push clean information to AI crawlers for e-commerce, use:

• Product and Offer Schema: Make sure prices, stock and variations are clearly defined.

• Review and FAQ Schema: Make verifiable evidence immediately clear.

LLMs still struggle with heavy client-side JavaScript when compared to HTML. This means that content which looks good on your site’s front end can be completely invisible to an AI crawler if it relies heavily on JavaScript rendering.

“A good trick that I like to use is to just test the view source and see if you can see all of the key content on the page,” Dena Warren explains. “If you can't see that when you're viewing the source, then AI is not going to be able to see it."

Just like traditional search engines, LLMs look beyond your site itself to determine its authority on a given topic. They review coverage and mentions on the wider web to decide whether you are trusted, relevant, and if your insights are validated by third parties. 

As well as highlighting the importance of active digital PR, this also makes careful monitoring and management of brand coverage essential to ensuring that your brand is correctly understood in AI search.

Rather than publishing self-serving listicles on your own domain, which is at risk of being labelled low-quality content, Leo Poitevin advocates for securing brand mentions in authoritative third-party industry publications. In this context, unlinked brand mentions are almost as beneficial for establishing AI visibility as traditional backlinks.

By establishing recurring mentions across listicles, articles and wider coverage on industry-relevant and reputable sites, LLMs will start to see your brand and content as a trusted source.

Moving away from traditional trust signals like link volume, LLMs perform sentiment analysis. They evaluate reviews and online mentions to build a broader consensus around your brand. For example, negative scores on platforms like Trustpilot could be flagged to warn users away from the brand.

To secure contextual user-generated content that matches long-tail queries, Dena Warren suggests crafting review request campaigns that ask leading questions like "Which cockpit size did you use this bag in?" or "How did this hold up in wet weather?"

Similarly, focus on building a strong brand presence on forums and in relevant communities to demonstrate trust through understanding and addressing your users’ pain points.

One of the biggest roadblocks to effective AI search optimisation today is the inability to accurately measure success, as traditional metrics like keyword rankings and organic impressions don’t apply. 

Leo Poitevin explains, "If you want perfection when you're doing GEO on LLMs, you're totally in the wrong way, because nobody's able to track everything... You will see traffic in analytics a bit, but the real money is on the brand searches, people searching on AI, seeing your brand, and coming to your site directly. You just have to trust the process."