The search results page you open today looks different from the one you opened two years ago. AI-generated summaries sit above the organic results, answering common questions without requiring a click. Perplexity and ChatGPT search are capturing a growing share of information-seeking queries. Google's own AI Overviews are eating into click-through rates for informational content.
If you depend on organic search traffic — and most websites do — these changes deserve serious attention. Not panic, but a clear-eyed reassessment of what SEO means in a world where the answer to the user's question is increasingly delivered before they ever see a list of results.
What Has Changed
Three structural changes are affecting organic search traffic in 2025, and understanding them is the starting point for any honest SEO strategy.
AI Overviews have reduced clicks for informational queries. When someone searches "how long to cook a whole chicken" or "what is a convertible note," Google now synthesises an answer directly in the results page. The user gets what they needed and often doesn't click anything. For content built around answering common factual or definitional questions, traffic has declined materially.
Zero-click searches are growing. Zero-click searches — where the user's query is resolved on the results page without a site visit — have been increasing for years, but AI Overviews have accelerated the trend significantly for informational content. Data from various analytics providers suggests that for broadly informational queries, click-through rates have dropped noticeably since AI Overviews rolled out at scale.
A new set of surfaces matters now. Being cited in an AI Overview, being referenced by Perplexity, or appearing in ChatGPT search responses now represents real visibility — separate from, and sometimes more valuable than, a traditional organic ranking. The game has acquired new playing fields.
Google AI Overviews Explained
Google's AI Overviews (formerly known during testing as Search Generative Experience, or SGE) are LLM-generated summaries that appear at the top of results pages for a wide range of queries. They draw from web content — including your site — to synthesise an answer, and they typically include citations to the sources used.
The nuance that many SEO commentators miss: being cited in an AI Overview is not necessarily bad for traffic. Studies have found that sites cited in AI Overviews can see increased brand awareness and, for certain query types, increased click-through from users who want more detail after reading the summary. The sites that lose most are those whose content existed purely to answer the question and had no other value proposition that would bring a user to the site.
Key insight: If your article's only job is to answer a question that AI can now answer directly, that article's traffic will decline. If your content also demonstrates expertise, provides unique data, or gives users a reason to engage with your brand beyond the answer, you retain value even in a zero-click world.
The Zero-Click Problem
The zero-click problem is real, but it's not new and it's not uniform. Zero-click has been growing since Google introduced featured snippets in 2014. What AI Overviews have done is expand the query types affected.
The queries most affected by zero-click are:
- Definitional queries ("what is X")
- Simple how-to queries ("how do I X")
- Factual lookups ("when did X happen")
- Comparison queries with simple, factual answers ("X vs Y for Z")
The queries least affected — and where organic clicks remain strong — are:
- High-intent transactional queries ("web design agency [city]")
- Navigational queries ("Desyn blog")
- Highly specific, complex, or opinionated queries
- Queries where the user needs to take action on a specific site
- Local search queries
This is useful information. It tells you to produce less purely informational content and more content that serves specific intent — content that requires your site to be part of the experience, not just a source of raw information.
What Still Works
Despite the legitimate changes happening in search, the foundational principles of SEO remain intact and in some cases have become more important.
Long-tail, specific, high-intent queries. These still drive clicks. When someone searches "web design for financial advisers London" or "how to write a homepage for a SaaS product targeting ops teams," they're expressing a specific need that requires specific expertise. AI Overviews appear less frequently for highly specific queries, and when they do appear, they rarely satisfy the full intent of a specific question.
Technical SEO fundamentals. Page speed, mobile optimisation, crawlability, and correct schema markup are unchanged in their importance. If anything, they matter more as Google decides what content is worth surfacing at all.
Backlinks and authority. The signal that a domain is genuinely authoritative in its field still matters. AI search systems — including Google's — weight content from recognised authorities more heavily. Building genuine authority through consistent, high-quality publication and earning links from relevant sources remains a durable strategy.
Optimising for AI Answers
Optimising for AI-generated answers — whether in Google's AI Overviews or in tools like Perplexity — is a newer discipline, but the principles are not mysterious.
AI systems synthesising answers favour content that is:
- Clearly structured — with descriptive headings, logical flow, and content that directly answers questions in the text beneath each heading
- Definitively stated — content that makes clear, direct claims rather than hedging everything is more useful as source material for a synthesised answer
- Schema-marked — FAQ schema, HowTo schema, and Article schema all help search engines understand what type of answer your content is providing
- Properly cited — linking out to authoritative sources and being linked to from authoritative sources signals reliability
- Authoritative in voice — content written by people with demonstrable expertise, and attributed to named authors with verifiable credentials where relevant
Notice that all of these are also good SEO practice for traditional search. Optimising for AI search and optimising for human-mediated search are increasingly the same activity.
The E-E-A-T Advantage
E-E-A-T — Experience, Expertise, Authoritativeness, and Trustworthiness — is Google's quality framework for evaluating content. It became a more prominent ranking consideration with Google's Helpful Content updates, and in the age of AI Overviews it has become even more central.
AI Overviews preferentially cite sources that demonstrate genuine expertise. A blog post written by a named practitioner who references their specific experience, includes original analysis, and is backed by a site with a coherent niche and a history of high-quality publication will consistently outperform a generic, authorless listicle — even if both rank in organic results.
The Helpful Content era has made one thing clear: Google is trying to surface content written for humans by humans with actual expertise. Every cynical SEO shortcut is a bet against that direction of travel.
Building E-E-A-T is a long-term project: it requires publishing under named authors, building an author profile with verifiable credentials, earning coverage and links from your industry, and maintaining consistent topical focus. It can't be faked quickly. That's exactly why it's worth building.
Our SEO Recommendations for 2025
Based on everything above, here is how we're thinking about SEO strategy for the sites we build and manage:
- Think about being cited, not just ranked. Structure your content to be excellent source material for an AI-generated answer. If an AI were going to summarise your topic, would it reach for your content first? Why?
- Invest in high-intent, specific content. The further down the purchase consideration funnel your content sits, the more immune it is to zero-click effects.
- Build genuine author authority. Named authors with real expertise and cross-platform presence signal E-E-A-T at a level that anonymous content cannot.
- Prioritise technical fundamentals. Core Web Vitals, schema markup, clean crawlability — these are not optional extras.
- Don't chase AI Overviews at the expense of depth. Content that is shallow enough to be fully resolved by an AI summary is content that no longer serves a strategic purpose. Add the layer of depth, original perspective, or specific application that an AI can't.
SEO in 2025 is more demanding than it used to be — not because the fundamentals have changed, but because the shortcuts have closed. The path to durable search visibility now runs through genuine expertise, genuine usefulness, and genuine authority. That was always the correct path. AI search has just made it the only one.