SEO has supposedly been dying for most of the time I have worked in it.
The proposed cause changes. Social media was going to replace search. Mobile apps would make websites irrelevant. Voice search was apparently waiting just around the corner. Now generative AI is expected to finish the job.
There is no doubt that AI assistants are changing how people find and interpret information. But changing search behaviour is not the same as replacing search engines. That distinction tends to disappear when somebody has a new acronym or monitoring platform to sell.
YouGov’s report, Searching for answers: How AI is changing online discovery in 2026, provides a more useful view of what is happening in Great Britain. Based on a survey of 2,003 British adults, it examines where people search, how AI fits into the process and what prevents them from trusting it more. (YouGov)
The findings do not show people abandoning search engines for chatbots. They show the search journey becoming more fragmented.
Search engines remain the default
Among British online searchers, 85% had used a search engine during the previous 30 days. AI assistants or chatbots were used by 31%.
AI is therefore a meaningful search channel, but it is not the dominant one. Maps and news websites were each used by 39% of respondents, while social media and video platforms were used by roughly a third. People are not moving neatly from one universal search interface to another. They are using different tools for different jobs.
The task-level figures make this clearer. Search engines were the leading starting point for every activity YouGov tested: 67% started there when asking a specific question, 64% when researching products and 60% when buying products. For troubleshooting, 56% began with a search engine, compared with 14% using video and 13% using an AI assistant.
AI had its clearest role in answering specific questions, where 21% used it as their starting point. That is significant. It is still some distance from replacing search engines.
The more useful conclusion is that one Google ranking now explains less of the customer journey. Someone may discover a company through search, ask an AI assistant to compare it with competitors, watch a demonstration on YouTube and then look for opinions on Reddit. The original ranking still mattered. It simply did not own the entire decision.
This is why search data needs to be treated as more than a channel report. Used properly, it can reveal the questions, comparisons and uncertainties shaping the market — something I discussed in more detail in how to turn search data into market intelligence. (Grumpy Old SEO)
AI is becoming part of the process
YouGov classifies 58% of British online searchers as “AI searchers” because they at least occasionally begin a query with an AI assistant.
That is a broad definition. Only 18% start at least one AI query each day, while another 16% do so several times per week. Twenty-four per cent use AI no more than once a week, and 35% never start a search that way.
AI search is established enough to matter, but it is not yet a universal habit.
Growth is also concentrated among people who already use it. Seventy-one per cent of AI searchers said their use had increased over the previous year. However, only 12% of non-users expected to use it more during the coming year, while 59% did not expect to start.
This suggests deeper adoption among existing users rather than an immediate wholesale conversion of everyone else. Businesses should therefore avoid both extremes: pretending AI is replacing everything, or dismissing it because conventional search still leads.
The most common reason people use an AI assistant is to get a direct answer, selected by 54% of AI searchers. But they also use it to summarise information, verify claims, compare options and troubleshoot problems.
Those are interpretation tasks. People are asking the system to reduce several pieces of information into something more manageable.
That changes the role content plays. A page must still be discoverable, but it also needs to contain information that is clear, specific and sufficiently supported to be reused in an answer. This does not require turning every page into a sequence of tiny questions and answers. It requires making the point properly and providing evidence for it.
AI is rarely the whole journey
Only 15% of YouGov’s AI searchers described an AI assistant as their usual starting point. Thirty-six per cent used it alongside other sources, while 21% used it after other forms of search. Another 24% used it occasionally for particular questions.
The generated answer was not always the end of the process either. Twenty-two per cent said they most often clicked links supplied by the assistant, while the same proportion compared the answer with other websites or applications. Twenty-four per cent usually stopped searching.
These are reported habits rather than observed click data, so they should not be mistaken for a reliable AI referral forecast. People are not always perfect witnesses to their own browsing behaviour.
Even so, the findings challenge the idea that every generated answer removes the need for further research. Sometimes the answer is sufficient. At other times it creates a need for verification, detail or an official source.
For a business, this means that being mentioned may have value without producing an immediate click. It may influence recognition, comparison or a later branded search. But a mention is not automatically an endorsement, and a citation is not the same as commercial impact.
Trust is still the problem
Search engines retain a substantial trust advantage. Seventy-seven per cent of British online searchers said they trusted search engine information at least somewhat. For AI assistants, the figure was 31%.
Trust rose to 49% among people who already used AI for search, suggesting that familiarity and confidence are connected. The study cannot tell us which causes the other.
When asked what would improve trust, AI users favoured official sources, direct source links and clearer explanations of how an answer had been reached. They also valued comparisons between several sources and visible indications of uncertainty.
These are not exotic requirements. People want to know where a claim came from, whether it is current and whether they can check it.
Businesses cannot control whether an AI system behaves responsibly. They can make their own information clearer and easier to verify. They can also examine whether their brand is being represented accurately, what evidence supports it and why competitors may be easier to retrieve or cite. That is the purpose of an AI Search Audit, rather than simply counting mentions across a collection of prompts. (Grumpy Old SEO)
What this means for SEO
Technical SEO still matters. Google’s own guidance says the same foundational SEO practices apply to AI Overviews and AI Mode. Pages must still be indexed and eligible for a conventional search snippet, and Google says there are no additional technical requirements or special AI markup needed to appear. (Google for Developers)
Crawling, indexing, internal linking and clear textual content have not become obsolete. They are simply not the whole strategy.
Rankings also provide an increasingly incomplete view of visibility. A business may rank well in Google but be poorly represented in AI-generated comparisons. It may appear in AI answers through third-party sources while receiving little direct referral traffic. Or it may be visible across both while being described inaccurately.
The starting point should therefore be the questions that matter during a real buying decision. What do customers need to know? Which alternatives do they compare? What risks do they investigate? Which sources do they trust?
Where those questions cut across technical SEO, content, positioning and AI visibility, the problem is strategic rather than channel-specific. That is the territory covered by strategic SEO consulting, not another isolated optimisation checklist. (Grumpy Old SEO)
SEO versus GEO is still the wrong argument
The YouGov research does not show search engines being replaced. It shows AI assistants joining the collection of tools people already use to find and interpret information.
Search engines remain the leading starting point for most tasks. AI use is growing, particularly among existing users, and it is becoming useful for answers, summaries, comparisons and verification.
Those facts are not contradictory.
SEO still supplies much of the technical and informational foundation. GEO, AIO or whichever term survives the current naming competition broadens the discussion to how brands are understood and represented in generated answers.
That is legitimate work. It is not a replacement for SEO, and it is not a reliable method for obtaining a stable number-one ranking in ChatGPT.
Search has become more fragmented and less tidy to measure. The sensible response is to understand where customers look, what evidence they need and whether the business is represented accurately across the journey.
Less dramatic than declaring SEO dead. More useful, though.
Sources
The primary source is YouGov’s Searching for answers: How AI is changing online discovery in 2026 — Great Britain. The survey was conducted online between 21 April and 29 May 2026. Its findings describe reported behaviour and attitudes rather than directly observed searches or clicks. (YouGov)
The discussion of Google AI Overviews and AI Mode is based on Google Search Central’s official guidance, AI features and your website. This describes Google’s systems and should not be treated as evidence for how independent products select or cite sources. (Google for Developers)