GEO: a practical guide to generative engine optimisation

Generative engine optimisation (GEO) is about increasing your visibility in AI-generated answers, such as Google’s AI Overviews and AI Mode, and ChatGPT search.

GEO became a hot topic as Google rolled out AI Overviews: in the UK from August 2024, and to more than 200 countries in May 2025. Since then we have seen a clear effect on search results, especially for informational queries. What many people miss is that GEO isn’t only about answering questions. It’s about becoming the source AI answers draw on.

• What GEO is
• Whether SEO still matters
• The difference between GEO and SEO
• How to optimise for GEO
• How to measure the results

• GEO is about visibility in AI answers: becoming the source that ChatGPT, Google’s AI Overviews and Bing cite.
• SEO is still the foundation: in one study, pages ranking first on Google appeared in AI answers 25 per cent of the time.
• The focus shifts from “the best page” to “the best answer”: AI wants clear, structured answers backed by sources.
• Fewer clicks, better clicks: traffic falls, but Google says the clicks it sends from results with AI Overviews are higher quality.
• Measure in Google Search Console: compare impressions with clicks for question-style queries.

Now let’s go through each part in detail.

The short answer: GEO is about becoming the source AI platforms refer to. They favour content that is clearly structured, up to date and from authoritative sources.

Some people will get their answer directly in Google and no longer click through to a result. Traffic is therefore likely to fall as the share of so-called zero-click searches keeps growing. This has been going on for a while, through featured snippets. For now it mainly affects informational queries.

Here’s what most people miss: AI answers aggregate facts from many sources, but they often borrow a distinctive perspective, analysis or conclusion from a single source.

An example: search Google for “How do AI Overviews affect SEO traffic?”. The AI answer summarises data from several sources, but when it comes to interpretation and recommendations, it often cites a single expert.

So the goal isn’t only to answer “what”, but to own the answer to “why” and “how should I think about this?”. This matters most for anyone who sells expertise. As a search marketing agency, we want to be the voice AI quotes when someone looks for strategic guidance, and the same goes for any consultancy or specialist firm.

Yes, very much so. Tomasz Rudzki, co-founder of ZipTie, studied 25,000 searches across ChatGPT, Perplexity and Google’s AI Overviews. As Search Engine Journal reported:

Good search engine optimisation is about being mentioned by relevant sources, keeping your content up to date and giving people good answers. That obviously matters for GEO too. Is it all the same? Of course not.

For now there is a strong correlation between ranking well in Google and appearing in AI Overviews. ChatGPT search is less transparent: OpenAI says its own crawler, OAI-SearchBot, is what surfaces websites in ChatGPT’s search features, but it doesn’t say how much it relies on Bing or other search providers. Either way, the relationship with classic rankings isn’t one-to-one.

Since AI Overviews arrived, we have seen concrete differences in how people search and in what gets shown.

People have started asking longer, more conversational questions. Instead of “GEO tips” we now see:

• “Compare the benefits of GEO for an online shop and a SaaS company”
• “How should I prioritise between SEO and GEO on a limited budget?”
• “Which GEO strategies work best for B2B companies in the UK?”

These are questions that need nuanced answers, not just a list of tips. Whatever market you work in, you’re probably already seeing this shift in Search Console.

Schema.org markup has always mattered for SEO, because it helps search engines understand what a page is about and who is behind it. But it is no shortcut into AI answers. Google says there is no special structured data you need to add to appear in AI Overviews or AI Mode, and that your markup should match the visible text on the page.

• FAQ content: clear questions and answers on the page. Since 2023 Google only shows FAQ rich results for well-known government and health websites, so for most businesses the value lies in the content rather than the markup.
• How-to content: step-by-step guides with numbered steps. Google retired How-to rich results in 2023, so write the steps out clearly instead of relying on markup.
• Article markup: to state the author, publication date and updates.

If you don’t have the technical skills in-house, an SEO agency with experience of structured data can help.

What I’m looking forward to most is that AI can offer a better interface and more tailored answers. If anything, that should increase the number of customers you can win through SEO.

SEO ranking factors have been tested far more thoroughly. But here are some I’ve seen work in practice and that other experts have confirmed:

• Answer the question clearly, ideally at the start of the paragraph. AI likes direct answers.
• Shift the focus of each page from “the best page” to “the best answer”.
• Use headings that ask a question, such as “How long does it take?”.
• Use bullet lists. AI answers often lift them straight in.
• Back up what you say with figures and sources. Facts build trust.
• Make it clear who is saying what, for example “we at [brand]” when the insight is your own.
• Show expertise (E-E-A-T). AI gives weight to credibility.
• Update often. Content that goes stale tends to lose ground.
• Cut the fluff. AI wants substance, not marketing waffle.

So much of this comes down to your on-page SEO and how you write your articles.

When we searched Google for how to measure the ROI of SEO, the AI Overview often cited one particular source that uses its own calculation model with concrete figures. Why that source?

• A clear formula: presented at the start of the section
• A concrete example: with real numbers, not “X%” as a placeholder
• Visual structure: formula, bullet list and table
• Fresh data: the article had been updated within the previous six months

This is what I mean by owning the perspective. They didn’t just give an answer. They gave the most useful answer, with a method of their own.

• Traffic will most likely fall, especially from informational queries.
• The number of customers who want to buy stays the same. So even if traffic falls, enquiries probably won’t be hit too hard, and your conversion rate should rise.
• What many people overlook is that Google’s biggest competitor probably isn’t Bing but other channels, such as TikTok, which younger people in particular use as a search engine, for recipes for example. Google and ChatGPT building a more appealing search experience should increase overall interest in search.

I think Google will come out of this ahead, even though OpenAI got off to a good start. Google has far greater resources and infrastructure. I’m also convinced that SEO and GEO together will matter more than ever.

Tools such as Semrush can track results much as they track traditional keywords. But most of the measurement will probably come from impressions and clicks in Google Search Console and Google Analytics. Bear in mind that Search Console doesn’t report AI Overviews or AI Mode separately: Google includes them in the overall figures for the Web search type.

1. Open the Performance report in Google Search Console

2. Add a filter on Query

3. Choose Custom (regex)

4. Enter: ^(what|how|why|which|when|who|where|can|should|does|is)

5. Click Apply

You now see all your question-style queries. Compare impressions with clicks. A rise in impressions without a matching rise in clicks can indicate that you’re appearing in AI Overviews: people get the answer without clicking.

Filter on your most important informational pages and compare click-through rate (CTR) between:

• a period before AI Overviews started appearing for your queries (in the UK they launched in August 2024)
• a comparable period afterwards

Search Console only keeps the last 16 months of data, so for a comparison that goes further back you need figures exported at the time.

A falling CTR on informational pages while visibility holds steady or rises can be a sign that AI answers are showing for those queries, whether or not you are the source they cite.

Semrush and Ahrefs both have tools for tracking visibility in AI answers, such as Semrush’s AI Visibility Toolkit and Ahrefs’ Brand Radar, which covers AI Overviews, AI Mode, ChatGPT and Perplexity.

In Google Analytics, compare the conversion rate of your organic traffic over time. If traffic falls but the conversion rate rises, that suggests your GEO work is paying off: you are getting fewer but better-qualified visitors.

• Carry on with SEO, broadly as before
• Focus on giving the best answer, not the longest page

If you’d like your business to be the source AI answers cite, take a look at our AI search services, or ask for a free search audit, where we also look at your potential in AI search.