Is AI killing SEO?

For more than two decades, SEO has run on a fairly stable logic: the user puts a question to Google, the search engine returns a list of results, and the user then chooses a website to find out more. Businesses therefore competed to appear in the highest positions, because a good ranking usually brought more visits and from there created sales opportunities.

AI is changing that very journey.

Rather than handing the user ten links and leaving them to read and pull it all together, Google AI Overviews, AI Mode and tools such as ChatGPT and Gemini are increasingly able to compile information and give an answer right there in the conversational interface. At Google I/O in May 2025, Google said AI Overviews had been used by more than 1.5 billion people across 200 countries and territories. Google also said that in large markets such as the United States and India, the group of queries carrying AI Overviews recorded more than 10% growth in search volume.

The question is therefore no longer theoretical: if users can get an answer without visiting a website, does SEO still have value?

When part of the clicks start to disappear

A Pew Research Center study shows this change can already be observed in real behaviour. Pew analysed the browsing data of 900 adults in the United States during March 2025, covering 68,879 Google queries in total. About 18% of the queries in the sample produced an AI summary.

What is notable lies in the behaviour after the user sees the AI answer. On results pages carrying an AI summary, users clicked a traditional search result on about 8% of visits, while the figure was 15% on pages without an AI summary. The rate of clicking straight through to the source cited inside the AI summary was only about 1%. Users were also more inclined to end the browsing session immediately after receiving the AI answer: 26% against 16% on a results page without an AI summary.

These figures point to a reality the SEO industry can hardly ignore: for certain kinds of query, particularly informational questions, AI can reduce the need to visit a website directly.

It would be too early, however, to conclude from that that AI is “killing SEO”.

Another study, by Semrush, covering more than 200,000 keywords during 2025, shows a more complicated picture. Keywords producing AI Overviews usually carry a higher zero-click rate, but when Semrush tracked the same group of keywords before and after AI Overview appeared, the zero-click rate actually fell from 33.75% to 31.53%. According to Semrush, this shows one cannot assume AI Overview is always the direct cause of a fall in clicks; the user’s search intent and the nature of the query still play a very large part.

In other words, AI is changing how traffic is distributed, but there is not yet enough basis to say that the need to search, or the role of the website, is disappearing.

SEO is not going away; the goal of SEO is widening

What is most notable is that Google itself does not regard AI search as a system separate from SEO.

In its official guidance updated in July 2026, Google states that the existing SEO principles continue to have value for AI Overviews and AI Mode, because these features still rest on Google Search’s indexing, ranking and quality assessment systems. Google also says AI responses use techniques such as retrieval-augmented generation to find and reference suitable web pages before forming an answer.

This changes how we should understand SEO.

Previously, success was usually measured by a fairly simple chain: keyword, ranking, click and traffic. In the era of AI search, businesses face a new layer of competition: whether their content is understood correctly by the search system, judged trustworthy enough and brought into the process of forming an answer.

A website may not receive the click it once did, but the brand can still appear in the AI answer. A piece of research can become a reference source. A product can be understood accurately by AI in terms of its features and who it is for. A business can be mentioned when a user asks AI to suggest a suitable supplier.

That is when SEO is no longer only a competition over ranking but widens into a competition over visibility.

This change is already large enough that Google has begun changing how it measures too. In June 2026, Google introduced a separate Generative AI report in Search Console, letting websites track how many times a URL appears in AI Overviews, AI Mode and other AI experiences in Search. That is a notable signal: beyond traditional clicks and rankings, how present you are within AI answers is becoming a metric that can be measured in its own right.

Mass-produced content will find it steadily harder to compete

AI is also changing the kind of content that has value for SEO.

When anyone can produce dozens of articles in a few minutes, content that merely compiles common knowledge becomes ever more of a commodity. An article of the “10 ways to do X” or “5 benefits of Y” kind will struggle to create a long-term advantage if it only repeats what is already on the internet.

In its latest guidance for AI search, Google recommends focusing on content with value of its own, based on experience, expertise and a real point of view; it also warns that producing large numbers of pages simply to cover every query variant does not create better quality and may breach the anti-spam policy.

That makes the things once sometimes regarded as SEO’s “side matters” more important: original research, case studies, real data, direct experience, the author’s expertise, brand standing and content AI cannot readily produce for itself simply by compiling existing sources.

SEO may therefore become harder, not simpler.

Does GEO replace SEO?

The appearance of terms such as GEO – generative engine optimisation – or AEO – answer engine optimisation – has led many to think an entirely new discipline is replacing SEO.

Google’s official position, however, is fairly clear: optimising to appear in AI search experiences is still regarded as part of SEO. Google also warns against chasing unverified “GEO tricks” such as creating special files just for AI, chopping content into small pieces mechanically, or seeking unnatural brand mentions.

That does not mean GEO is a meaningless idea. It is still useful for describing a new layer of problem: how to get a brand and its content to be present within the answers AI generates. But in essence, SEO and AI search optimisation have a great deal of overlap.

Websites still need to be crawled and indexed. Content still needs to be useful. Company information still needs to be clear. Standing still has to be built. User experience still matters. What is new is that these factors now serve not only a results page of ten links but also AI systems compiling answers from many sources.

AI is not killing SEO, but it is killing an old way of doing SEO

To my mind, this is the most appropriate way of looking at it at present.

AI can reduce part of the traffic a website once received from simple queries. That is real and has begun to appear in the data. But at the same time, AI search still needs the internet, still needs websites and still needs sources of sufficient quality to form its answers.

So SEO will not disappear. What is losing value is the way of doing SEO that revolves only around producing a lot of content, covering a lot of keywords and waiting for traffic.

Over the coming years, businesses will probably have to look at SEO more broadly. Not only what position the website holds on Google, but whether the brand appears in the new search journey; whether AI understands the business correctly; whether the content is valuable enough to be used as a reference source; and whether, when the customer has to decide, the business is among those considered.

If SEO was once mainly the game of “being found”, AI is adding several new layers:

Being found → Being understood correctly → Being trusted → Being mentioned → Being chosen.

SEO is not dead.

But the SEO of the era that chased only rankings and traffic is entering a period of drastic change.