Is SEO Dead in the Age of AI?

SEO has been declared dead roughly every three years since 2005. This time something real did change: the click. What did not change is who gets cited.

Is SEO Dead in the Age of AI? — Troiana insight cover

In short

No, but it has changed shape. Google still handles the overwhelming majority of searches and its results still send more traffic than any other source; AI Overviews and AI Mode have reduced clicks on informational queries, sometimes sharply, while the pages they cite are drawn from the same index by largely the same signals. AI assistants such as ChatGPT and Perplexity retrieve from the web and cite sources, which makes being crawlable, authoritative and clearly structured more valuable, not less. What died is the strategy of ranking generic pages for traffic; what survives is being the source worth citing.

What actually changed

Three things, and they are worth separating.

The click became optional. AI Overviews, and AI Mode where it is available, answer many informational queries on the results page. Studies through 2025 measured organic click-through falling by a third or more on affected queries. The search still happened; the visit often did not. This is a genuine loss for sites whose business was informational traffic, and it is the fact behind every "SEO is dead" headline.

Assistants became a search channel. A meaningful and growing share of research now happens in ChatGPT, Claude, Perplexity and Gemini. These systems retrieve from the live web for anything current or specific, and they cite what they retrieve. The sources they choose are decided by signals that look a lot like the ones search engines use: authority, clarity, relevance, freshness, and whether the page is fetchable at all.

Query length grew. People ask assistants and AI Mode full questions, with context. Intent is stated more explicitly than it was in a three-word search, which makes pages that answer specific questions more findable, and pages that target vague head terms less so.

What did not change: Google is still where most searches start, the index that AI Overviews draw from is the same index, and the pages that get cited are the pages that would have ranked.

The parts of SEO that matter more now

Being the source. An AI answer is a summary of a few pages with citations. The value has moved from being on the page of ten links to being one of the three sources the answer names. That is won by the same things that always won rankings, applied harder: first-hand data, specific expertise, clear structure, a named author with a track record. Our guide to getting cited covers the practicalities.

Technical accessibility. Assistants fetch pages in real time. A page that is slow, blocked by robots rules, hidden behind JavaScript rendering, or missing from the index cannot be cited. Crawlability is now a visibility decision, not just a search one.

Brand. When the click disappears, the name in the citation is what the searcher retains. Brand searches, direct visits and mentions across the web are the signals that both search engines and assistants use to decide who is worth citing, and they are the traffic that survives a zero-click answer.

Specific questions. Long, explicit queries reward pages that answer exactly one thing exactly. The search intent discipline matters more, not less.

Commercial and local queries. AI answers a definition; it cannot deliver a plumber or complete a purchase. Queries with a transaction at the end still send the click, and the pages that win them are pricing pages, service pages and product pages built to convert.

The parts that matter less

Generic informational traffic. The "what is X" article that existed to rank and show an ad has lost most of its reason to exist. If AI can answer it from the top three sources, the tenth version brings nothing.

Volume. Publishing a hundred adequate pages was a strategy when each one could take a slice of long-tail traffic. Now it is scaled content abuse by Google's definition and invisible to assistants that cite the best source rather than all sources.

Keyword mechanics. Exact-match phrases, density, keyword-led URLs. These were already fading; systems that understand meaning made them irrelevant.

Traffic as the metric. Sessions were the number SEO reported. When a growing share of visibility produces no session, the metric undercounts the value. Citations, brand search volume, share of AI answers and assisted conversions are the replacements, imperfect as they are.

What a strategy looks like now

  1. Decide which questions you are the best answer to, and answer them with something nobody else has: your data, your cases, your position.
  2. Publish fewer, better pages with a real author, a clear structure, and the answer in the first paragraph.
  3. Make the site trivially fetchable: fast, indexable, open to the crawlers you want, with structured data where it helps.
  4. Build the brand off-site: mentions, reviews, partnerships, being named in the places assistants read.
  5. Own the transactional pages: pricing, services, comparisons, the queries that still click.
  6. Measure visibility, not just sessions: citations, branded search, AI referral traffic.

This is what the discipline is being called generative engine optimisation, or GEO, and the complete guide goes deeper. It is not a replacement for SEO. It is SEO with the click removed from the middle of the equation, and the honest practitioners were already doing most of it.

The one-sentence answer

SEO as a traffic-harvesting tactic is in decline; SEO as the practice of being the source that search engines and assistants trust is more valuable than it has ever been, because the reward for being cited is now the whole answer rather than a tenth of a results page. If you would like to know where a site stands under the new rules, book a call and we will audit it for both.

Common questions

Is SEO still worth it in 2026?

Yes, with a changed emphasis. Google still handles most searches and sends more traffic than any other source, and AI Overviews and assistants cite pages drawn from the same index by similar signals. What has stopped paying is generic informational content published for traffic; what pays is being the authoritative, fetchable source on specific questions, and owning the commercial queries that still produce clicks.

How have AI Overviews affected website traffic?

On informational queries where an Overview appears, organic click-through has fallen substantially, with published studies through 2025 reporting drops of a third or more for top positions. Commercial, local and navigational queries are far less affected. Sites cited within the Overview recover part of the loss; uncited sites bear all of it.

Will ChatGPT replace Google search?

Not on current evidence. Assistants have taken a meaningful and growing share of research and question-answering, but Google's query volume remains vastly larger and it has integrated AI answers into its own results. The realistic picture is several search surfaces, all retrieving from the web and all rewarding the same qualities in a source.

What is the difference between SEO and GEO?

SEO optimises for ranking in a list of links; generative engine optimisation optimises for being cited in an AI-generated answer. The underlying work overlaps heavily: authority, clarity, structure, crawlability and original information. GEO adds emphasis on being fetchable by AI crawlers, answering specific questions directly, and measuring citations rather than clicks.

What should I do if my organic traffic is falling because of AI?

Check which queries lost clicks; if they are informational, that is the AI Overview effect and not a penalty. Shift effort toward pages that still click, such as pricing, services and comparisons; make your best content citation-worthy with original data and clear structure; ensure AI crawlers can fetch your site; and start measuring brand search and citations alongside sessions.

Have something worth building right?