Have you ever asked ChatGPT to recommend a product or service? You’re part of a small but growing minority if you have.
Estimates of the volume of search-like queries put through LLMs like ChatGPT vary, but are thought to be around 2-10% of the total number of searches, including those made through Google and other traditional search engines.
2%? Why bother with AI search at all?
So, while most of your customers are still defaulting to traditional search, there are two reasons why you should still optimise your content to be recommended within LLMs:
- Futureproofing is never a bad idea. Let’s take the more conservative figure of 2%. It may not sound like a lot, but taking 2% of a ginormous market when your technology didn’t even exist five years ago is staggering. Which leads to point 2 …
- LLM search will catch on. Yes, on the one hand, humans are creatures of habit and we’re used to Googling stuff. On the other hand, LLMs offer a level of personalisation that a traditional search can’t even begin to match. You can Google “half-term holiday ideas for family” but you can’t get it to tailor a precise response to a detailed query that includes your kids’ ages, preferences and specific needs. An LLM can do that effortlessly.
How do LLMs choose their sources?
You can watch my recent short video for a detailed explanation of how LLMs search for information to include in their responses. In a nutshell, they don’t just search the words that people use: they add contextual terms to provide as comprehensive an answer as possible. That means optimising for AI search is less about specific keywords and more about meeting intent (i.e. understanding what people mean when they search).
Optimising for AI search: where to start
As we mentioned in our 2026 predictions piece, SEO is shifting to what we’re calling “Everything Optimisation”. That means thinking more holistically about your online brand presence.
The good news is, there is no new set of magic tricks you need to learn to optimise your content for inclusion in AI search results. Our go-to phrase here at NSDesign is “Good SEO is Good GEO” (i.e. ‘Generative Engine Optimisation). All the things you should already be doing to optimise your website for Google also apply to optimising it for LLMs. These include:
- Clear delineation of topics within your website
- Accessibility best practices
- Meeting user intent
- Strong technical SEO including structured meta data
- Engaging, easy-to-read content
- Citations and ‘brand proof’ from other websites
However, if you’ve not paid too much attention to the last one on this list, now is the time to get serious about your online PR. LLMs draw heavily from third party platforms like Reddit, news sites, Wikipedia and so on. Only optimising your website won’t get you featured in LLMs: they’ve got to be able to build a picture of your business from around the web.
Kicking off your PR drive
You need a PR strategy that makes your brand visible on social media sites, industry sites, communities, news sites and more. Most small businesses can’t be everywhere, so think about what’s most natural and logical for you. You might not realistically get your own Wikipedia page, but could you ask customers to write you a TrustPilot review or share your social media posts?
A good starting point is to run a few test queries through an LLM such as ChatGPT and see if you can spot any patterns in the sources it recommends. If they’re all from news sites, that tells you that you need to think of a way to get press coverage. If it’s product review sites, that’s where you need to be. And so on.
AI search is still a new concept and it’s not possible to say exactly what logic LLMs use when they select one source over another. We can, however, try to learn from what we’ve seen so far and update our strategies for the best chance of visibility in AI search.
For more help with your 2026 digital marketing and AI strategies, get booked onto one of our digital skills masterclasses, or talk to our team.