SEO vs. AEO: How the customer journey has split in 2026

The same user who searches on Google in the morning asks ChatGPT for a recommendation in the afternoon. Two different paths to a purchasing decision. And most companies are still only optimising for the first one.

This is not a theoretical problem. It is a measurable one, and it is growing every day.

What is changing right now, and why it’s different this time

Every few years, a new trend sweeps through the world of SEO: mobile-first, voice search, featured snippets. Most of these haven’t changed the core process: users search, Google lists the results, users click. This time, it’s different.

AI Overviews now appear in up to 47 per cent of all Google searches. Pages that rank in first place lose an average of 34.5 per cent of their click-through rate as soon as an AI Overview appears above them. Among B2B SaaS companies, 40 per cent report measurable declines in organic traffic. These are not outliers; this is the new normal.

What has changed structurally is that users stop searching sooner. Not because they haven’t found an answer, but because they already have one. Google provides it directly on the search results page. ChatGPT, Perplexity and Copilot do the same thing, outside of Google. According to Gartner, by the end of 2026 around 25 per cent of organic search traffic will have shifted to AI chatbots and voice assistants. This isn’t just noise. It’s a structural shift.

The two funnels and why they require a different approach

Both funnels lead to the same goal: awareness, consideration, decision. But the path between them works in a fundamentally different way, and that changes what you need to do to be visible.

Awareness

SEO: The user types in keywords, sees a list of links, clicks on several results and compares them.

AEO: The user asks the AI a question, receives a single synthesised answer, sees two or three cited sources, and nothing else.

Consideration

SEO: Users open between 5 and 10 tabs, read various sources, compare them manually and form their own opinion.

AEO: The AI has already carried out the comparison; the answer includes a recommendation straight away – no need to switch tabs, no manual comparison required.

Decision

SEO: The user returns to the best source, remembers the brand or adds the page to their bookmarks, and converts on the website.

AEO: Users act directly on the basis of the AI recommendation, click on the cited source – if at all – or make a purchase without ever having visited the website.

The key difference

With SEO, the user decides who to trust. With AEO, the AI decides who to cite.

That sounds like a minor detail. But it isn’t, because the implication is radical: in the AEO funnel, there is no second page, no second link alongside it, no second chance. Anyone who doesn’t appear in the response simply doesn’t exist at that moment.

And here’s the figure that most people aren’t yet aware of: 60 per cent of citations in Google AI Overviews come from URLs that don’t even rank in the top 20 organically. This means that a good ranking doesn’t automatically protect you from being invisible in the AEO funnel, and a poor ranking doesn’t rule you out. AEO follows a different logic. If you don’t understand this, you’ll be optimising for the wrong target.

What really determines which sources AI models cite

AI models do not rank pages. They extract answers. The difference may sound technical, but it has direct practical implications for anyone who produces content.

78 per cent of all AI-generated answers contain list formats. Articles cited in AI Overviews contain, on average, 62 per cent more specific facts than those not cited. Among all markup types, the FAQ schema has the highest citation rate in AI-generated answers. According to early GEO studies, pages with FAQ markup are included in AI Overviews 3.2 times more frequently than pages without structured Q&A markup.

The reason behind this: AI models have been trained on high-quality, structured data. According to the latest figures, Wikipedia accounts for 47.9 per cent of all ChatGPT citations. The FAQ schema mimics precisely this format. The models recognise such structures as reliable sources of answers. Unstructured running text, which only gets to the actual answer after three paragraphs, is skipped – not because the content is of poorer quality, but because it is more difficult to extract.

The second signal that most people underestimate: brands are cited 6.5 times more frequently in AI responses when the reference comes from third-party sources rather than from the AI’s own domain. Perplexity provides an average of 13 brand mentions per query, drawing primarily from Reddit, G2, PCMag and Gartner. ChatGPT favours market leaders and well-known names. Those without strong coverage from third-party sources rarely feature in ChatGPT’s responses, no matter how good their own website is.

What you can actually do

Structure content in such a way that AI can extract it

The most common mistake: articles that only get to the point after a long introduction. AI models won’t wait. They look for the direct answer and either find it straight away or move on. Every relevant page needs a concise answer at the start (40 to 80 words), followed by FAQ sections, comparison sections and clearly defined sections. Correct implementation of the FAQ schema increases the likelihood of the page being cited by AI by over 36 per cent.

Systematically build up a presence on third-party platforms

PR and link building have recently played a largely supplementary role in modern SEO. In the AEO funnel, they are once again taking centre stage. When there are 6.5 times more citations from external sources than from one’s own domain, one’s own content alone is simply not enough. Guest articles in specialist media, expert interviews, listings in business directories and an active presence on platforms such as Reddit and LinkedIn – these are the sources from which ChatGPT, Perplexity and Gemini build their answers. Anyone who isn’t featured there has a structural visibility problem that no on-page fix can solve.

Your own data as a magnet for citations

In 2026, synthesised content will be interchangeable. What AI models favour as sources: content containing quantified facts, specific data and clear timestamps. Your own studies, internal analyses and survey results are the sort of content that others do not have and which are therefore cited. One in-house survey per quarter is worth twenty generic articles.

Introduce new metrics

By 2026, rankings and organic traffic will only represent part of the picture. Anyone who does not also measure how often they are cited in ChatGPT, Perplexity, Google AI Mode and Gemini is navigating in the dark. The relevant metrics are now known as ‘Share of Model’ (how often do you appear when AI discusses your category), ‘Citation Rate’ (the proportion of relevant AI responses in which you appear) and ‘Competitive Share’ (your mentions versus those of your competitors). Tools such as AEO Vision, Gauge and Profound now make this monitoring possible across all platforms.

The conclusion

SEO and AEO are not an either/or situation. They serve the same customer in two fundamentally different ways. In the SEO funnel, the user carries out the research themselves, compares options and makes a decision. In the AEO funnel, the AI handles the research and decides which options to suggest.

The key point is this: if you only focus on one funnel, you’ll be invisible in the other. And the other funnel is growing, with up to 47 per cent visibility in Google searches and millions of daily queries to ChatGPT, Perplexity and the like.

Can you be seen in both?

Frequently asked questions about SEO and AEO