INSIGHTS How do you optimise a website for AI?

In this article, Kerri Fear, The Escape’s Head of Performance, delves into the traditional SEO tactics and new AI-specific optimisations that work together to optimise websites for AI visibility.

AIO, AEO, GEO, LLMO, AISEO, RAG - What does it all mean?

If you’ve been on LinkedIn recently, you’ll recognise some of these acronyms. But what do they all mean, and what’s it got to do with SEO and AI?

• AIO - Artificial Intelligence Optimisation
• AEO - Answer Engine Optimisation 
• GEO - Generative Engine Optimisation 
• LLMO - Large Language Model Optimisation 
• AISEO - Artificial Intelligence Search Engine Optimisation 
• RAG - Retrieval Augmented Generation

Now, if you are au fait with the language around AI and SEO, you’re probably thinking,  “aren’t these all the same thing?” And you would be right, these are all terms describing essentially the same thing – optimising your website to be found for AI tools.

At The Escape, we settled with GEO - Generative Engine Optimisation, as for us it’s the most accurate description of what we are working to achieve. 

What is Generative Engine Optimisation?

In the simplest format, it’s the act of optimising a website to be found easily by generative AI tools, and to encourage them to utilise the content in their response, along with a link back to the website (a citation).

Sound familiar? It’s similar to Search Engine Optimisation (SEO) - with the goal moving from getting clicks from search engines like Google & Bing to being referenced by AI tools such as ChatGPT, Perplexity & Copilot.

How does Generative AI work?

It would be daft to dive into how we optimise for a tool without giving you an idea of how it works. Most generative AI tools work in a very similar way, using ChatGPT as an example:

1. The user prompt is received
    
2. The query is processed to understand the intent and to identify missing information from the tools' LLM database. This is done by creating query fans of relevant, semantic terms.
    
3. The tool launches a real-time website search, which will start with a search engine like Google or Bing.
    
4. The tool selects several ‘trusted’ sources to analyse for its response to the user.
    
5. It summarises the information found in its database and via live website retrieval.
    
6. The tool uses its LLM (Large Language Model) to turn the information summarised into a coherent response by predicting the most probable next word in a sentence.

Most of the responses you receive from generative AI tools are purely processed from their LLM, which is linked to a huge database of information that the tool was trained on. Such as:

• Books
• Websites like Wikipedia
• Documents
• Human feedback from other users

This is why the response you receive may not be accurate or correct initially, as it’s created from pattern recognition and potentially outdated information. These tools also run live retrievals through the web, which is where websites need to be able to provide the right information, in the right format, for these tools to discover and use the content in responses.

Isn’t this just SEO for AI tools?

Generative AI tools are still in their infancy, despite their incredible capabilities for users. Which means they are still developing how they retrieve information.

A huge number of SEOs believe “It’s just SEO”, so you carry on rolling out all the tried and tested methods that improve rankings. It’s not quite as simple as that - don’t get me wrong, I don’t believe “SEO is dead” - another popular phrase making the rounds on LinkedIn.

Mike King, keynote speaker at October’s Brighton SEO, says it best:
‘SEO isn’t dead. SEO is deprecated.’

The rise of generative AI means users have shifted how they search. This has changed the entire universe of search, meaning the ‘old way’ is also shifting.

We still need traditional SEO, as huge numbers of users are still using traditional search engines. But it will slowly become more and more irrelevant as users migrate to easier and more accurate ways of getting what they want.

What do you do when something you do is deprecated? You don’t abandon it immediately - you maintain what you have, upgrade where you can and investigate new solutions.

In the SEO universe, the new solution is GEO.

How do you optimise a website for AI?

Our team has conducted a huge amount of research into AI optimisations, and continues to, as this is an ever-changing industry. What we know so far is as follows.

SEO activities that support GEO

While generative AI is still developing its data retrieval systems, some classic SEO tasks work incredibly well.
 

1. Structured Data / Schema Markup

Adding this to your website gives bots/crawlers easy-to-digest information about your page, which Gen AI loves to digest and use.

2. Well-formatted content

Content written with hierarchical heading structures, in easy-to-read sections with reference links, is a recipe for success.

3. High-quality, helpful content

Google’s been craving this for years, because it’s what users really want from the results. So, of course, Generative AI also wants to gobble up good content, too!

4. Original data

Real, first-party data from your business, with the citations to back it up, is gold for both SEO & GEO. It shows expertise & authority, elevating your content & brand trustworthiness.

5. A fast website with no errors

Users don’t like slow websites with broken pages, nor do Gen AI and search engine crawlers/bots. Do the housekeeping and keep your site speedy & clean.

6. A robots.txt file that doesn’t block AI crawlers

This may be a no-brainer, but blocking AI crawlers in your robots.txt file means they aren’t going to be crawling your site or referencing you. If your website uses CloudFlare, request that it unblock the top Gen AI tools, as these are blocked by default currently.

7. Avoid duplicate / rehashed content

Unique content is key for search and Gen AI tools. Don’t duplicate content across the site without canonicalising it, and avoid copying content from other sites to your own! Write content that is unique to your website & brand.

8. Do not use AI-written content verbatim

You can use Gen AI to support your content writing, but avoid using only AI content on your website. Bring value with human perspectives, thought-leadership and first-party data to make your content unique!

10 ways to improve visibility on Generative AI

Now, this is what you’re really here for, isn’t it? The list of what you can do to improve your website for generative AI.

1. Chunk your content

Chop your page and article content into quick & easy-to-digest chunks.

Just check out your ChatGPT responses and Google’s AI Overviews to see how it showcases information. Align your content in a similar format.

2. Create query fans for your target keywords

One of the most interesting things I learnt at Brighton SEO was how Gen AI tools worked. They don’t find information for a single keyword; they break the keyword into around 10 semantically-relevant terms and find information for each of these.

These ‘query fans’ are the reason AI responses are so well-rounded, with details you may not have specifically asked about!

3. Add citations and references

If you’re making a statement, back it up with context & proof. First-party data will set you apart from your competitors in an AI world.

4. Add summaries to long pages and articles

If you have a long guide or a page with huge amounts of information, create a short, concise summary that Gen AI can utilise.

5. Avoid the fluff

If you’re giving the answer to a question, keep it simple and concise. Gen AI wants to provide the quickest, easiest & most accurate answer for users. So you should too!

Bad news if you run a recipe site - Gen AI doesn’t care for the backstory of how you learnt how to cook mince pies at your nanas in the 90s.

6. Satisfy the whole user intent

Moving away from a single keyword approach to content, towards satisfying the whole intent of a user is key.

This also feeds into the semantic query fans, as a user looking for ‘winter boots’ is likely to want to know ‘how long these boots last in winter’, ‘does leather go mouldy?’, ‘best socks for boots’ and where to buy them. You can see how quickly you can create more pages to satisfy these related queries.

7. Improve brand perception, mentions & reviews

If you imagine Gen AI bots as if they were humans doing research on an item before purchasing it, where do they look to check if your brand is legitimate?

Gen AI checks these too - so make sure your Trustpilot, Google & Glassdoor reviews are excellent. And that your brand is mentioned by users on sites like Reddit, IMDB & Wikipedia are positive! The density of good brand mentions online is key to showing trust.

8. Avoid JavaScript

Google and other search engines can crawl and understand content rendered by JavaScript - most Gen AI and LLMs do not!

Most of them are built to avoid rendering this kind of content, so don’t accidentally hide everything by using JavaScript!

9. Don’t forget about non-text content

When we talk about content, it’s easy to assume it’s all textual.

Diversify your content by incorporating your own unique images, videos, charts, graphs & diagrams. Perplexity.ai is reported to be more likely to cite sources that include multiple types of content.

10. Stay up to date

An article you wrote three years ago may still rank 1st on search engines, but Gen AI retrieves data live - so it expects and favours the most up-to-date content.

Make sure to utilise last-updated Schema, datePublished and dateModified along with changelogs for long-form content.

How do you track GEO?

At the time of posting, most Gen AI tools do not openly provide data on how often a website is cited in responses. It’s rumoured to be in the works with ChatGPT, and Perplexity offers better in-chat ‘Source visibility’ for users. There are a number of third-party tools that claim to track term & brand visibility alongside citations; however, there is little clarity on how these tools get their data.

We can manually report on citations, competitors, URLs used and overall visibility in the responses by running query tests on the platforms themselves. Over time, this regular insight can be used to see long-term improvement in citation volume.

The only true data points we can collect on Gen AI visibility are the number of users who have visited your website from Gen AI tools, by checking Google Analytics referrals. If this number grows, and the pages being visited are the ones that have been optimised, it’s working!

The Escape’s approach to GEO

Just like traditional SEO, our work with GEO starts with detailed research to understand where your brand currently stands in this environment. 

We’ve created a new generative AI discovery audit (a GEO audit for short) to diagnose the technical and content fixes required to improve the visibility of your website in AI. In this audit, we also benchmark key query groups to understand how your brand and website are referenced against your competitors, to provide real, actionable insights for improvement.

Start optimising for AI today

I hope that these tips have helped bring GEO out of the shadows and into the light for you. The performance team loves talking about SEO & AI, so if you don’t know how to start or want to pick our brains, book a chat with us today.