If your rankings look fine but your clicks are sliding, you are not imagining it. Google now answers a huge share of searches before the user ever reaches a blue link, and the click that used to be yours is being absorbed at the top of the page. This post gives you the real benchmark numbers, explains why no single “average” tells the whole story, and walks through how to measure and protect your own traffic.

The short answer: how much traffic are sites losing?

Here is the direct answer, before the nuance: model 15% to 60% click loss on affected informational queries, and use these planning assumptions for the rest of your SEO forecast.

The headline figure most teams now plan around is a 61% organic CTR decline on queries where an AI Overview is present, with paid CTR falling 68% on those same queries. Both come from Seer Interactive’s analysis of 3,119 informational queries across 42 organizations, where organic CTR fell from 1.76% to 0.61% between mid-2024 and late 2025. Ahrefs, using a separate 300,000-keyword dataset, found that an AI Overview correlates with a 58% lower click-through rate for the page sitting in position one, up sharply from the 34.5% the same team measured in early 2025.

Different studies, different samples, same direction. That tells you something important: this is a structural shift, not a measurement quirk.

Average AI Overview traffic loss benchmarks

If you only take one section away, take this one. These are the working benchmarks to plan against, drawn from the most-cited independent research.

Two important things to keep in mind, and you should know these before you end up sharing those frightening stats everywhere.

First, those 61% and 58% figures don’t necessarily mean your whole website lost 60% of its traffic – which, let’s be honest, sounds like a nightmare. What they’re actually talking about is a very specific situation, specifically when an AI Overview does kick in – that’s a big difference.

And second, that 35% increase in citations is actually a correlation, not a proven cause of anything. And the researchers themselves make it clear that it’s more like a trust signal before it’s a traffic source.

Why there is no single “average” traffic loss number

Anyone quoting you one tidy percentage is selling certainty that does not exist. Your actual loss depends on:

• Query intent, if you’re answering questions that people just want info on, you’re going to get hit hard, but if it’s a transactional query or one where people are searching for a brand, you’re likely to come out okay.
• Does an AIO even show up for that keyword in the first place? If it doesn’t appear for some reason, then you’re not going to lose traffic to it.
• What else is happening? Sometimes there are so many stacked up, a featured snippet jumps out, People Also Ask kicks in and then an AI Overview pops up too, and before you know it the whole screen is crowded out by all that and your website doesn’t even get a look in.
• Where you were ranked before? If you were already at the top of the list, then you’ve got more to lose when the AI Overview arrives.
• Brand authority, generally, well known brands are more likely to keep their traffic.
• Are you being cited or buried? Being named as part of the Overview is a very different thing from being ranked below it.
• How much you need to dig? If people are just looking for quick answers, it’s going to be bad news for you, but if they’re looking for something in-depth, they might still find you.

This is why one study might find a 34.5% drop in traffic, another a 58%, and another a 61% – none of these are wrong, they just measure different things in different ways.

What AI Overviews are doing to organic CTR

AI Overviews cut clicks for one blunt reason: they answer the question before the user reaches your result. When the summary sits at the top of the page and resolves the query, there is no reason to scroll.

That is most dangerous for “what is” questions, simple definitions, basic comparisons, and easy how-to searches, exactly the content that used to drive heavy top-of-funnel traffic. The damage shows up even when your ranking does not move. A page can hold position one, lose nothing in the SERP, and still bleed a third or more of its clicks because Google placed a generated answer above it.

And the floor keeps dropping. Ahrefs’ position-one CTR loss climbed from 34.5% to 58% in roughly eighteen months as the novelty wore off and users learned to trust the summary. Pew Research adds the most sobering number of all: only about 1% of users click a source link inside an AI Overview. Google reads your content, summarizes it, presents it as its own answer, and sends almost no one back.

What ChatGPT is doing to website traffic

ChatGPT is a different animal. It is less a direct “traffic thief” and more a replacement for the search itself. When someone asks ChatGPT to define a term or compare two options, that search never reaches Google, and never reaches you.

What zero-click Google elements are doing to organic traffic

Zero-click search is not new. AI just made it more aggressive. Long before AI Overviews, Google was resolving queries on the results page through:

• Featured snippets
• Knowledge panels
• People Also Ask
• Local packs
• Shopping modules
• Direct answers
• Video carousels

AI Overviews are simply the newest and hungriest member of that family, and they often appear stacked with the older ones. When a snippet, a People Also Ask box, and an AI Overview all fire on the same query, the entire above-the-fold SERP becomes a wall of answers and your link is buried beneath it. The cumulative effect is what is pushing whole categories, retail, news, and marketing among them, into year-over-year traffic declines.

The queries most and least at risk

If your content lives in the at-risk column, treat it as exposed and act first. This is not a guess. Roughly 88% of the queries that trigger AI Overviews are informational, the educational, explanatory, comparison-style searches that AI summarizes most cleanly. By contrast, commercial queries make up only around 9% of triggers. The more your page reads like a quick-answer definition, the more replaceable it is.

The queries that hold value do so because a generic summary cannot replace them. The pattern is consistent: the deeper the consideration and the more original the substance, the safer the click. AI Overviews punish shallow answers and reward source-level authority.

How to calculate your own AI traffic loss

Industry benchmarks tell you the weather. This tells you whether your roof is leaking. Run this in an afternoon:

1. Export your queries from Google Search Console (last 16 months).
2. Segment by intent, informational, commercial, brand, navigational.
3. Identify which queries show an AI Overview, manually or with a rank tracker that flags the SERP feature.
4. Compare CTR before and after the AI Overview first appeared for each query.
5. Separate ranking loss from feature loss, if your position dropped, that is a different problem.
6. Model by query group, not by domain. A site-wide average will hide where the bleeding actually is.

Holding ranking constant is the whole point. It isolates the click you lost to the Overview from the click you lost to a competitor outranking you. Those need completely different fixes.

How to protect organic traffic from AI Overviews and zero-click search

You cannot opt out of the new SERP. You can become the source it cites. These are the tactics that move the needle:

• Create original data. A number nobody else has is a number AI has to attribute to you.
• Add genuine expert commentary. AI systems lean toward content with named expert input when choosing what to cite.
• Build tools and calculators. A working tool cannot be summarized into a paragraph.
• Use clean comparison tables. Structured, scannable formatting is what AI extracts most easily.
• Strengthen author and entity signals. Clear bios, credentials, and consistent naming help AI trust the source.
• Add schema markup. Article, FAQPage, HowTo, and Person schema give crawlers a cleaner read.
• Answer the query fast, then go deeper. Lead with the direct answer in 40–60 words, then earn the deeper read.
• Refresh aggressively. Freshness studies show recently updated content earns several times more AI citations than stale pages, content updated within roughly 30 days has been found to attract about 3.2x more citations.
• Build branded demand so people search and click you by name, the one query AI cannot intercept.
• Earn citations from trusted sites that AI already references.
• Track AI visibility, not just rankings.

The new SEO KPI stack

Clicks are no longer the only scoreboard. If you measure only organic clicks, your dashboard will tell you you are failing even while your brand grows inside AI answers. Track the full stack:

• Organic clicks
• Organic CTR
• AI Overview presence on your target queries
• AI citation frequency
• Brand mentions inside AI tools
• Assisted conversions
• Direct traffic lift
• Newsletter signups
• Returning users
• Revenue per organic session

When AI references your brand among only three or four sources, you have earned something a single click never gave you: trust at the moment of decision.

Is SEO still worth it?

Yes! Of course. It will always be valuable. It's adapting and we must adapt with it. That means adapting to how weak informational SEO getting crushed. Pretending otherwise will cost you a year you do not have. The thin "what is" post built to rank and collect AdSense pennies is the exact content AI eats first.

The winners in this cycle will not be the sites with the most ranked URLs. They will be the brands that become cited entities, the names AI systems reach for when they assemble an answer. You have to be relentless about becoming the cited source, not just another blue link. That is the digital real estate worth building.