Has AI Overview Led to Traffic Decline? What the Data Shows
By Ben, Founder
AI Overviews do suppress organic click-through, but the impact is uneven. Informational queries with an AI Overview panel present lose between 30 and 80 percent of clicks depending on placement below the panel. Branded, transactional, and comparison queries see far smaller drops. The practical implication: segment your keyword portfolio by intent, then decide which clusters to defend with traditional ranking and which to target for AI citation instead.
You opened Google Search Console, saw clicks falling, and you can’t tell if AI Overview is the cause or an algorithm update or just seasonality. Worse, impressions are up while clicks are down, and nobody told you what that signal means. This article gives you the numbers, then a way to read your own data so you know which of your keywords are actually bleeding. For the wider picture, here’s how AI is reshaping SEO broadly.
The traffic numbers are real, and worse for some sites than others
The decline is not in your head. A Guardian study found that sites previously ranked first can lose up to 79% of their traffic when their result appears below the AI Overview panel. That is not a rounding error. That is most of your clicks gone.
Bain reports that 80% of users now rely on zero-click results, meaning they get what they need without visiting a single page. The Mail, citing Bain data, found a two-thirds drop in click-through rate when an AI Overview appears. Two-thirds.
Here is the part that confuses people. BrightEdge data shows impressions rising while CTR falls. Your content is being seen more, clicked less. That pattern tells you the cause is structural, not seasonal. Seasonality moves impressions and clicks together. This moves them apart.
One caveat matters. These headline numbers come from news and publishing verticals, where AI Overviews hit hardest. Brand and SaaS SEO data tells a different story, and that difference is the whole point of the next section.
Why informational queries take the biggest hit
The reason is simple. An AI Overview absorbs definitional and how-to queries entirely. Someone types “what is a content pillar” or “how to set up GSC,” reads the panel, and never clicks. The answer was the destination. There was nothing left to click for. Here’s how AI Overviews work and which queries they absorb in more detail.
Now compare that to a branded query. When someone types your company name, they want you, not a summary of you. Those searches don’t trigger AI Overviews the same way, and even when they do, the user still clicks through to the source they were already looking for.
Transactional and comparison queries hold up too. “Best X for small teams” or “X pricing” require trust signals an AI Overview can’t provide. People want to see the actual page, the actual price, the actual reviews before they spend money. The panel can’t close that gap.
This is the diagnostic framework most pages skip. Andy’s live SERP audit data, fetched in real-time for each keyword research run, shows this intent-based CTR segmentation consistently across client keyword sets. Informational queries with an AI Overview present lose the most clicks; branded and transactional queries hold up significantly better. Everything starts by the search intent, and the intent decides how much you lose.
How to diagnose your own traffic drop in Google Search Console
Stop guessing. Your GSC data already holds the answer if you read it by intent instead of in aggregate. Here is the order I’d work in.
- Filter by query type. Pull your informational queries, the “what is,” “how to,” and definitional ones. If impressions are up and clicks are down on that group, that is the AI Overview suppression signal. Clear and specific.
- Compare click rates before and after the May 2024 AIO rollout for the same query clusters. Same keywords, two time windows. If the cluster fell off a cliff after the rollout, you have your culprit.
- Export queries with high impression share and sub-1% CTR. These are your AIO-absorbed keywords. They get shown, they don’t get clicked, and they are unlikely to recover with a title-tag tweak.
- Segment branded versus non-branded queries separately before you draw any conclusion about total traffic health. Total traffic is an average, and the average hides the truth. Your branded clicks may be fine while your informational clicks collapse.
Once you have that export, you’ll want to know which keyword types consistently trigger AI Overview panels so you can predict the pattern instead of reacting to it. The keywords that trigger a panel are the keywords you measure differently from now on.
The strategic response: defend what’s defensible, get cited in the rest
You don’t fight every battle. You pick the ones you can win. Split your keywords into two buckets and treat each one differently.
Informational keywords with an AI Overview present: change the goal. Stop trying to rank a blue link below the panel where 79% of clicks already vanished. Aim to be the source cited inside the Overview instead. That means original data, a clear stated opinion, and content the model wants to quote.
Branded and transactional keywords: keep investing in traditional ranking signals. These clicks survive. Your homepage, your pricing page, your comparison pages, your product pages. That is where money is made, and the panel isn’t taking it.
This is the Reforge defensibility frame applied to a content portfolio. Commodity informational content is not worth defending against an AI Overview, because you can’t out-rank a feature Google built into the results page. As we hold it: if you do not have a strong opinion, your content is going to be replaced by AI, because AI can easily create and generate content. The fix is not more generic articles. It’s defensible ones.
Doing this by hand across hundreds of keywords is slow. Andy’s keyword audit workflow flags which clusters fall into each bucket automatically during the research run, using live SERP data plus the brand interview collected from your own website crawl and onboarding session. You get a list of articles that you want to write and a list of articles that you do not want to write, with the reason attached to each.
What this means for your content roadmap going forward
The roadmap changes in four concrete ways. None of them are “write more, faster.”
First, stop writing generic informational content that an AI Overview will absorb without citing your source. A “what is X” explainer that ranks below the panel is work you do for Google’s summary, not for your traffic. Prune it, don’t publish it.
Second, shift toward defensible content. First-party data, your brand’s strong opinions, original research, real case studies. This is your content and your strong opinion, and it’s the only informational content worth investing in now, because it’s the only kind a model can’t generate without you.
Third, track branded search volume alongside click volume. When people start searching your name, that’s demand you built, and an AI Overview can’t intercept it. It’s a leading indicator of SEO health that survives the zero-click shift.
Fourth, count LLM citations as ranking. A mention in ChatGPT or Perplexity is the new rank, even when it doesn’t produce a Google click. Most cited sources don’t even rank in Google’s top 20, so if you measure only blue-link clicks you’ll miss the visibility you’re actually earning. For the full picture, here are the right metrics to track once clicks stop being the primary signal.
FAQ
How much traffic can AI Overviews take from a top-ranked page?
A lot. The Guardian found that pages previously ranked first can lose up to 79% of their traffic when the result sits below the AI Overview panel. Bain data, reported by The Mail, shows roughly a two-thirds drop in click-through rate when an AI Overview appears at all.
Which types of content lose the most traffic to AI Overviews?
Informational and definitional queries, by a wide margin. “What is” and “how to” searches get answered inside the panel, so the click never happens. Branded, transactional, and comparison queries hold up because they need trust signals or a specific destination the panel can’t supply.
Is SEO dead or evolving in 2026?
Evolving, not dead. The KPI moves. Branded search volume and LLM citation count matter more than raw click count now. Most cited sources don’t even rank in Google’s top 20, so measuring only blue-link clicks tells you less every quarter.
How do I tell if AI Overviews are causing my traffic drop in Google Search Console?
Look for impressions up and clicks down on your informational queries specifically. That split is the AI Overview suppression signal. Then compare the same query clusters before and after the May 2024 rollout, and segment branded from non-branded before you judge your total traffic.
Is AI Overview bad for my website?
It depends on your keyword intent. It’s bad for commodity informational pages, because the panel answers the question for the user. It’s neutral or even positive for branded and transactional pages, where clicks survive and your real conversions live. Diagnose which bucket your keywords fall into, then act per bucket.
