The decade-long playbook of rank, click, convert is breaking down in real time. Here is what the numbers say and what is replacing it.
Let’s start with the thing that is actually happening, rather than the thing marketers and SEO professionals would prefer to believe.
In February 2024, Gartner predicted that traditional search engine volume would drop 25% by 2026, with AI chatbots and virtual agents taking market share from Google. At the time, many industry commentators pushed back. The prediction seemed extreme. Bing had just added an AI chatbot and gained essentially zero market share. Why would this be different?
Then the data started coming in.
What the Data Shows
A Pew Research Center study tracked 68,000 real search queries and measured what users actually did. When Google’s AI Overviews appeared in the results, users clicked on a website 8% of the time. Without an AI Overview present, they clicked 15% of the time. That is a 47% relative reduction in click-through rate from a single change to the search results page.
Seer Interactive ran a longer-term analysis, tracking 3,119 informational queries across 42 organisations from June 2024 through September 2025. Organic click-through rates for queries with AI Overviews dropped from 1.76% to 0.61% over that period, a 61% decline. Paid search fared worse: click-through rates dropped from 19.7% to 6.34% for the same query set.
The counterargument has been that queries without AI Overviews are a safe harbour. The data does not support this either. Seer found that even those queries, where no AI Overview appears, lost 41% of their click-through rate year-over-year. Something broader is shifting, not just the presence of a single feature on the search page.
The cumulative impact on publishers is measurable. Chartbeat data shows a 33% decline in Google search traffic globally between November 2024 and November 2025. HubSpot saw its monthly organic visits fall from approximately 13.5 million in November 2024 to under 7 million by December of that year, continuing to around 6 million by early 2025. The primary driver was Google’s December 2024 algorithm update, which penalised content disconnected from a site’s core expertise. HubSpot had built traffic on topics like resignation letter templates and real estate licence guides, which had nothing to do with its CRM platform. When Google tightened topical authority requirements, that traffic evaporated. The company’s CEO separately acknowledged the industry-wide trend on an earnings call, noting that organic search traffic is “declining globally” and that AI Overviews are “giving answers and fewer people are clicking through.” Those are two distinct problems reinforcing each other.
Business Insider lost 55% of its organic search traffic over a three-year period ending April 2025 and cut 21% of its staff. These are not fringe publishers with weak content strategies. They are organisations that built competitive advantages on exactly the type of high-volume content that search engine optimisation rewarded for years.
Sixty percent of Google searches in 2025 now end without any click to a website. That figure was around 54% in 2017. The trend has been building for years but AI Overviews accelerated it sharply.
Where the Searches Are Going
The obvious question is: if people are not clicking through to websites, where are they getting their information?
Some portion stays on Google. AI Overviews answer the question directly on the results page, which is precisely the problem for publishers. But a growing share is going somewhere else entirely before Google is even consulted.
ChatGPT now has over 800 million weekly active users and ranks among the five most visited websites globally. Perplexity processed 780 million queries in May 2025 alone, a 239% increase from the same month the previous year. AI platforms overall grew from 0.02% of global internet traffic in 2024 to 0.15% in 2025, a more than seven-fold increase in one year.
An important caveat is necessary here. 0.15% is still a small share. Google remains structurally dominant. Research from September 2025 found that Google still sends approximately 345 times more referral traffic to publishers than all AI platforms combined. Search is not dead. The behaviour around it is changing.
What the data describes is not a collapse of search but a fragmentation of information retrieval. Users are increasingly reaching AI tools first for informational and research queries and reserving search for transactional and navigational intent. The queries where traditional search optimisation has always driven the most traffic, the “how does X work” and “what is the best Y” type of question, are precisely the queries that AI handles most effectively. Commercial and local search shows more resilience. The high-funnel informational query, which underpinned content marketing as a discipline for the past decade, is the category under the most pressure.
The Nomenclature Problem
The industry’s response has been, predictably, to invent acronyms.
GEO (Generative Engine Optimisation), AEO (Answer Engine Optimisation) and AIO (AI Optimisation) are proliferating faster than consensus on what they mean. Different tools and consultancies use them interchangeably or with subtly different definitions. The naming is unsettled because the field itself is unsettled.
What all of them point toward is the same underlying shift: optimising for visibility within AI-generated responses rather than ranking position on a traditional search results page. The goal is no longer to be the link that a user clicks. It is to be the source that the AI cites when it generates an answer.
This is a meaningful change in what the work actually involves.
What Actually Works in AI Search
Research into how AI platforms select their sources has produced some consistent findings, though the landscape is volatile. Between 40% and 60% of cited sources change from month to month across major AI platforms.
A September 2025 paper from researchers analysing AI search systems at scale found that AI platforms show a systematic bias toward earned media, meaning third-party authoritative sources, over brand-owned content. Your website is not, by itself, your best asset for being cited. What other credible sources say about you matters more than what you say about yourself.
The signals that drive citations are in many ways an amplification of what good content strategy has always required: genuine expertise, third-party validation, clear structure and cited evidence. What has changed is the weighting and the mechanism. A few patterns that have emerged consistently:
Structure matters differently. AI Overviews average around 169 words and include approximately seven links when expanded. They strongly favour list-based formatting: 78% of AI Overview responses feature ordered or unordered lists. Content written in a way that makes it easy to extract a clean, quotable answer is more likely to be cited than dense prose, regardless of how well it ranks.
Schema markup signals intent to machines. Content with proper schema markup shows 30 to 40% higher visibility in AI-generated answers, according to research from Princeton and Georgia Tech. This is not new advice but it has more direct commercial relevance now than it did when it primarily affected rich snippets.
Earned authority from third parties. Reviews on G2 or Capterra, mentions in industry publications, citations in news coverage and discussions on Reddit and LinkedIn are the signals AI systems appear to weight heavily because they represent independent validation. Only 7.2% of domains appear in both Google AI Overviews and large language model results and those that do tend to have strong third-party citation profiles.
Being cited in AI Overviews compounds. Seer’s analysis found that brands cited in AI Overviews earned 35% more organic and 91% more paid clicks than brands not cited. The citation creates visibility that then drives additional traffic through traditional channels. The two are not strictly competing.
What Is Actually Dead
Traditional search optimisation is not uniformly dead. It is dead in a specific way.
The model that worked for the past decade, producing high-volume informational content, ranking for long-tail terms and converting the traffic, has broken down as a reliable growth channel for informational queries. The top-of-funnel content machine that agencies built their businesses around, that drove HubSpot to millions of monthly visitors and made content marketing a credible growth strategy, is no longer operating under the same conditions.
Transactional search is more resilient. Someone searching for a local plumber or a specific product to purchase is still largely clicking through to a website. Branded search is more resilient. What is not resilient is the middle layer: the how-to guides, the explainer articles and the “best X for Y” round-ups that AI now handles acceptably well on the results page itself.
For businesses that built their digital presence primarily on that layer, the adjustment required is significant. The playbook of publishing two articles a week optimised for informational terms and letting traffic compound over time will produce diminishing returns in a search environment where 60% of queries end without a click and where AI platforms increasingly satisfy informational intent before the user ever reaches a website.
The replacement is not a clean swap. It involves becoming the source that AI systems reference rather than the destination that users visit. It means building genuine domain authority in the way that earns third-party mentions. It means accepting that impressions and citations are now meaningful metrics alongside clicks and sessions.
A Realistic Assessment
The people selling AI optimisation as a turnkey replacement for search optimisation deserve some scepticism. The field is young enough that a lot of what is presented as best practice is better described as educated hypothesis. The 40 to 60% monthly variation in which sources AI platforms cite suggests the landscape has not stabilised. Investing heavily in optimising for a moving target, based on incomplete data about how citation decisions are made, carries real risk.
What is not speculative is the direction. Traffic patterns, zero-click rates, AI platform growth and publisher revenue data are all pointing the same way. The question is not whether the shift is happening but how far it goes and at what pace.
The most defensible position for any organisation right now is to stop treating search optimisation and AI optimisation as separate disciplines and to focus on what has always been the actual foundation of both: content that earns genuine authority because it says something true and useful that a specific audience could not easily find elsewhere. That kind of content travels in AI citations for the same reason it accumulated backlinks. It is worth referencing.
The keyword stuffing, the thin how-to articles and the content produced primarily to rank rather than to answer, that really is dead. It was already losing effectiveness before AI Overviews. The shift has simply accelerated the inevitable.