Why Is Your Website Losing Traffic as AI Replaces Google Search?

UK websites that consistently ranked well on Google are seeing organic traffic fall, even when their positions have not changed. The cause is not a penalty or an algorithm update. It is a structural shift in how people find information online: and it is accelerating in 2026.

AI search tools are answering questions that would previously have driven someone to click through to a website. The result is a shrinking pool of clicks even from pages that still hold top Google positions. Understanding what is happening: and what to do about it. It is now a practical business decision, not a theoretical one.

What Has Changed Since 2024

Several things have converged in the last 18 months:

• Google AI Overviews Google's AI-generated summaries appear at the top of results for a growing share of queries, particularly informational ones. The summary answers the question directly. Many users do not scroll to organic results below it.
• ChatGPT and Perplexity A growing number of users, particularly in professional and technical roles, no longer start research on Google. They ask AI assistants directly. These tools synthesise answers from multiple sources. The user never visits the underlying websites.
• Zero-click searches Even before AI Overviews, Google was answering an increasing proportion of queries within the results page through featured snippets, knowledge panels, People Also Ask boxes, and direct answers. AI Overviews extend this trend significantly.

Semrush data published in early 2025 found that AI Overviews appeared in approximately 13 percent of Google searches in the US. Ahrefs found that pages featured in AI Overviews received fewer organic clicks than those that ranked below the AI block. This effect (visibility without traffic) is the defining characteristic of the current environment.

What Your GSC Data Is Telling You

If your Google Search Console data shows impressions holding steady or rising while clicks are falling, this is the pattern. Impressions count every time your page appears in results, including when it appears below an AI Overview that already answered the question. Clicks only count when someone actually visits your URL.

A widening gap between impressions and clicks, combined with falling CTR, is the signature of the AI Overview effect. This is not something you have done wrong. It is a change in the results page itself.

Check this directly in GSC:

1. Go to Performance > Search Results
2. Enable all four metrics: Clicks, Impressions, CTR, Average Position
3. Compare the last three months to the same period last year
4. Look for impressions that have held or grown while clicks have fallen and CTR has dropped

If you see that pattern, AI Overviews and zero-click search are the most likely cause. If you see impressions falling alongside clicks, that is a different problem: a ranking issue that requires a different response.

The Traffic Channels That Are Replacing Google Clicks

As Google click volume falls for some query types, traffic is not disappearing entirely. It is moving to different channels and different behaviours. The businesses that hold their ground are building presence across multiple routes to discovery:

• Brand search When someone has already heard of you and searches your name directly, AI Overviews rarely intercept. Brand-driven traffic is more resistant to the AI effect than generic informational queries.
• Direct traffic Users who bookmark your site or navigate directly are not affected by changes in Google results. Building a reason for users to return directly (newsletter, account login, regular publication) creates a traffic base Google cannot take away.
• Email Owned channels are immune to algorithm changes. A weekly email to subscribers delivers traffic that does not depend on Google.
• LinkedIn and specialist communities For B2B businesses, LinkedIn remains a significant traffic and lead source. Unlike Google, the algorithmic dynamics are different and the audience is often more commercially qualified.

What Generative Engine Optimisation (GEO) Means in Practice

If AI tools are synthesising answers from multiple sources without sending traffic to them, getting cited in those answers becomes the new objective. The emerging discipline for this is Generative Engine Optimisation (GEO): structuring content so that AI systems cite it when answering related questions.

The research on what influences AI citation is still developing. What is clear from early studies and practical testing:

• Factual, specific content with sources AI systems prefer content that states facts clearly, attributes data to named sources, and avoids vague generalisations. A page that says "studies show X" is less citable than one that says "A 2024 Ahrefs study of 300,000 queries found X".
• Entity clarity Make it explicit who you are, what you do, where you are based, and who you serve. This builds the factual entity profile that AI systems use to represent your business in answers.
• Structured content formats FAQ sections, step-by-step instructions, numbered lists, and comparison tables are all formats that AI tools extract and reuse. If your content is structured, it is more citable than long prose.
• Brand mentions on third-party sites Being mentioned, linked to, or quoted on authoritative third-party sites trains AI models to associate your brand with relevant topics. This is traditional PR reframed for a generative world.

The Pages Most Affected and What to Do About Them

Not all content types are affected equally. The pages most likely to have lost traffic to AI are:

• Definition pages ("What is X?")
• List pages ("Best tools for X")
• How-to guides for simple tasks
• Comparison articles ("X vs Y")

These are exactly the query types where an AI system can produce a usable answer from existing web content without requiring the user to click anywhere. If your site relied heavily on these formats for traffic, the impact will be disproportionate.

Pages that retain value are those where the content cannot easily be summarised:

• Original research, data, or case studies from your own experience
• Detailed technical guides requiring hands-on knowledge
• Location or business-specific service pages
• Content requiring professional judgement rather than general information

The strategic response is to shift content investment away from informational pages that AI can answer and towards content that requires your specific expertise, real-world data, or a commercial relationship to produce.

Technical Changes That Help

Beyond content strategy, there are technical adjustments that can improve how AI systems and Google treat your content:

• Schema markup FAQ, HowTo, Article, and Organization schema help Google and AI systems understand the structure and authorship of your content. Pages with schema are more likely to have content extracted into AI Overviews with a citation link back to your site.
• Author pages and bylines Content with a named, credentialed author performs better in AI citation contexts than anonymous articles. Adding author bio pages with LinkedIn links and role descriptions is a straightforward implementation.
• Internal linking A well-linked site helps Google and AI crawlers understand which pages are most authoritative on each topic. Pages that receive many internal links from relevant content are more likely to be treated as the definitive source.
• Page speed and Core Web Vitals These remain Google ranking factors. A slow site that ranks at position 3 will lose proportionally more traffic to an AI Overview than a fast site that ranks at position 1 with a higher CTR.

Measuring Whether Your Strategy Is Working

Standard GA4 organic traffic figures will continue to fall in some segments regardless of what you do, because the clickable audience is genuinely smaller. Measuring success requires a different lens.

Track these alongside organic traffic:

• Brand search volume Rising brand queries mean your non-Google presence is working. Use GSC filtered to your brand name.
• Direct traffic Growth in direct visits indicates you are building return-user behaviour.
• GSC impressions for cited pages If a page gets cited in AI Overviews, you may see impressions rise even as clicks fall. This confirms the AI interception is happening, and the citation itself has value.
• Conversion rate by channel If organic traffic volume falls but conversion rate from those visits rises (because less-intent traffic was eliminated), revenue impact may be smaller than the traffic figure suggests.

What Not to Do

The AI search shift is causing some businesses to react in ways that create new problems rather than solving the original one:

• Publishing more thin content at higher volume AI-generated content at scale does not resolve an AI citation problem. It competes with the content causing the problem and typically makes on-site quality worse.
• Abandoning SEO entirely Google still processes billions of queries daily. Commercial intent queries ("buy X", "agency for Y", "service near me") remain high-value and less affected by AI Overviews than informational ones. Abandoning SEO in these segments would be a significant commercial error.
• Optimising for position 1 at the expense of all other signals Being ranked at position 1 below an AI Overview delivers fewer clicks than it did in 2023. Optimising title tags and meta descriptions for CTR is now more valuable relative to position than it was before.

The Businesses Most at Risk

Some business models are more exposed than others:

• Affiliate and comparison sites These rely on informational or comparison queries that AI systems can answer directly. The model is under structural pressure.
• Content-only publishers Sites that generate revenue through advertising on informational pages are seeing CPM fall as fewer users arrive from search.
• B2B businesses whose sales process starts with research If prospects previously found you through informational search, they may now be finding general answers from AI and forming brand preferences before they ever search for specific suppliers. This moves the sales funnel earlier, not later, which has implications for where to invest in brand building.

The businesses least affected are those with a distinctive brand, a specific expertise, or a local service relationship that AI systems cannot easily replicate. If your value is in execution rather than information, the structural change is less threatening.