Small Sites Lost 60% of Search Traffic. Big Brands Lost 22%.

Short answer: Search traffic has not fallen evenly. Over two years, sites drawing 1,000–10,000 daily pageviews lost about 60% of their search referrals, while sites above 100,000 daily pageviews lost about 22% — a 38-point gap between the small and the large, measured on the same platform in the same window. The decline tracks brand strength far more closely than it tracks site quality, which means the defensive move is not better optimization but a more recognizable name.

Key takeaways

• Chartbeat data across thousands of sites shows search referrals down 60% for small publishers, 47% for mid-sized, and 22% for large ones over two years.
• Google Search pageviews fell 34% year over year and Google Discover fell 15%, but total weekly pageviews fell only 6% — the traffic moved, it did not evaporate.
• When an AI Overview appears, branded queries gained CTR (+18.68%) while non-branded queries lost it (−19.98%). The split, not the average, is the story.
• Across 75,000 brands, branded web mentions correlated most strongly with appearing in AI Overviews (0.664) — well ahead of Domain Rating (0.326) and backlinks (0.218).
• AI chatbots send less than 1% of publisher referrals even after ChatGPT referrals grew 200%. Nothing is replacing the lost clicks at volume.

How much search traffic have sites actually lost?

The honest answer depends entirely on how big you already were.

Analytics firm Chartbeat, whose data covers thousands of publishers globally, broke two years of search referral change down by site size. The result is one of the widest performance gaps the search industry has produced:

Source: Chartbeat data reported by Search Engine Land, March 2026.

A small site lost close to three times what a large one did. Both were subject to the same algorithm, the same AI Overviews rollout, and the same zero-click behaviour.

The aggregate figures hide this completely. Across all sizes, the same dataset shows Google Search pageviews down 34% year over year and Google Discover down 15%. The Reuters Institute's Journalism, Media, and Technology Trends and Predictions 2026, drawing on Chartbeat figures for more than 2,500 news sites, reported a comparable 33% global drop for the year to November 2025, with Google Discover down 21%. That report also surveyed 280 media leaders across 51 countries, who collectively expect search referrals to fall a further 43% over the next three years.

The regional spread is nearly as wide as the size spread. The same Reuters Institute analysis put the US decline at 38% and Europe's at 17%.

Why did small sites lose nearly three times as much?

Because AI Overviews do not reduce clicks uniformly. They reduce clicks on queries where the user does not already have a destination in mind.

Marketing firm Amsive analysed 700,000 keywords across 10 websites and five industries — finance, education, SaaS, healthcare and pets — and separated the results by query type. The findings split cleanly:

Branded search did not merely survive the AI Overview. It performed better with one present.

That single row explains the size gap. A large site earns a substantial share of its traffic from people who already know its name and are navigating, not researching. A small site earns almost all of its traffic from people who typed a question and took whatever came back. AI Overviews answer the question. They do not answer the navigation.

Publishers describe the same mechanism from the inside. In a Press Gazette report on how newsrooms are managing the decline, the Daily Mail characterised its overall traffic impact as small, attributing that to the dominance of branded search in its mix. Bauer reported a single-digit year-on-year decline — far below the 33% headline figure. The Telegraph described itself as largely unaffected outside specific algorithm updates. These are not sites that optimized better. They are sites people ask for by name.

The same pattern appears in who gets cited

If branded demand protects clicks, it also predicts visibility. Ahrefs analysed 75,000 brands to find which signals correlate with being mentioned in AI Overviews, using Spearman correlation:

The classic authority metrics land in the bottom half. What correlates is how often the open web says your name.

The quartile breakdown is starker than the coefficients. Ranked by branded web mentions, the median brand in the top 25% recorded 169 AI Overview mentions. The quartile immediately below it recorded 14. The one below that, 3. The bottom quartile, 0.

That is not a gentle gradient. It is a cliff, and small sites are standing at the bottom of it.

Is this really AI Overviews, or is something else going on?

This is where most coverage of the topic overreaches, so it is worth being precise: the size gap is well measured, but the cause is not cleanly established.

Three complications deserve stating plainly.

First, Chartbeat did not attribute the decline to AI Overviews. The dataset shows search referrals falling. It does not isolate AI Overviews as the mechanism. Discover algorithm changes, ranking updates, and shifting news cycles all sit inside the same window.

Second, total traffic barely moved. Across the same publishers, total weekly pageviews fell only about 6% from 2024 to 2025. Search collapsed; the overall audience did not. Traffic shifted toward direct, internal and messaging channels. A site that lost 60% of its search referrals and 6% of its total audience has a channel problem, not an audience problem — and those call for different responses.

Third, Google disputes the premise. In an August 2025 post, Google stated that "total organic click volume from Google Search to websites has been relatively stable year-over-year" and that average click quality had improved. Google published no data, charts or Search Console figures alongside the claim, and independent measurement has repeatedly contradicted it. But note that Google's claim and Chartbeat's are not strictly incompatible: aggregate click volume across the entire web can hold roughly steady while clicks redistribute violently away from small sites and toward large ones. That redistribution is exactly what the size breakdown shows.

The defensible conclusion is narrower than the headlines and more useful: whatever combination of forces is at work, its effects concentrate on undifferentiated, non-branded, informational content — and that is what small sites mostly publish.

Where did the traffic go, if not to AI chatbots?

Not to the chatbots, at least not yet.

Even after ChatGPT referrals grew more than 200%, AI chatbots collectively account for less than 1% of publisher pageview referrals. The lost search clicks are not arriving from a new source under a different name. They are, in large part, simply not happening — the answer was consumed on the results page.

This is the part that makes "just optimize for AI search" incomplete advice. Optimizing to be cited in AI answers is worth doing, but it is not a traffic replacement strategy at current volumes. It is a presence strategy: you are buying the mention, the association, and the branded search that may follow later. We covered how that machinery works in AI Search Visibility.

What actually protects a site?

The data points consistently in one direction, and it is not toward better on-page optimization.

Build the branded demand. Branded web mentions were the strongest correlate of AI Overview presence (0.664), and branded queries were the only category that gained CTR when an AI Overview appeared. Everything else in the dataset is downstream of that.

Publish what a summary cannot substitute for. The Press Gazette reporting found that evergreen explainers and how-to content were hit hardest, while content requiring genuine expertise, judgement, opinion, or first-hand testing held its click-through. If a paragraph can replace your page, it will.

Stop reporting on sessions alone. A site that lost 60% of search referrals and 6% of total pageviews has been told something specific by its own analytics. Track branded search volume, direct traffic, mention rate and share of voice alongside sessions — see AI Search KPIs for a working set.

Confirm you are actually reachable. Being cited requires being crawled by the AI agents doing the citing, and robots.txt rules written years ago routinely block them by accident. This is the cheapest thing on the list to check and the most embarrassing to get wrong.

Earn the citation deliberately. Being mentioned is a repeatable process, not luck — How to Get Cited by AI sets out the mechanics.

Frequently asked questions

Did small websites really lose 60% of their search traffic?

Sites averaging 1,000–10,000 daily pageviews lost about 60% of their search referral traffic over a two-year period, according to Chartbeat data covering thousands of sites and reported in March 2026. Mid-sized sites lost 47% and large sites lost 22%. The figure is a median across the category, so individual results vary widely.

Are AI Overviews the cause of the traffic decline?

Not provably, and the underlying data does not claim so. Chartbeat measured search referrals falling without isolating a cause, and Discover changes, ranking updates and news cycles overlap the same window. What is well established is that when an AI Overview does appear, non-branded queries lose about 20% of their click-through rate while branded queries gain about 19%.

Why are large brands less affected?

Because much of their search traffic is navigational. People search for them by name rather than by question, and AI Overviews do not intercept navigational intent — branded queries actually saw CTR rise 18.68% when an AI Overview appeared. Large publishers including the Daily Mail and Bauer have attributed their comparatively mild declines to exactly this.

Is AI search traffic replacing the lost clicks?

No. AI chatbots account for less than 1% of publisher referral traffic even after ChatGPT referrals grew more than 200%. AI visibility is worth pursuing for brand presence and the branded demand it creates downstream, but at current volumes it does not substitute for lost search clicks.

Does domain authority still help with AI visibility?

Much less than most teams assume. Across 75,000 brands, Domain Rating correlated 0.326 with AI Overview mentions and backlinks correlated 0.218, while branded web mentions correlated 0.664. Authority metrics still matter for ranking, but how often the web mentions your name matters more for being cited.

What should a small site do first?

Separate the two problems. Check whether total audience fell as much as search traffic did — across this dataset, total pageviews fell only 6% while search fell 34%, which points to a channel problem rather than an audience one. Then verify AI crawlers can reach you, and shift measurement toward branded demand.

Where to start

Two checks worth running before you change any strategy:

• Crawler Checker — confirm GPTBot, ClaudeBot, PerplexityBot and Google-Extended can actually reach your pages. A blocked crawler makes every other effort here moot.
• Meta Tags Checker — verify the title, description and structured data that answer engines read when deciding how to describe you.

Both are free and take under a minute. If you want the full picture of where your brand currently surfaces across AI answer engines, the report demo shows what that measurement looks like in practice.