If you opened Google Search Console this month and saw impressions climbing while clicks fell off a cliff, you are living through the most common SEO complaint of the past two years. Site owners describe it the same way everywhere. Impressions up by hundreds of thousands, traffic down by half. Rankings stable, revenue sliding. The chart looks like an open crocodile mouth, and nobody on the team can explain why, because every traditional SEO metric says things are fine.
Here is the short version. Your traffic dropped after AI Overviews because Google now answers a large share of informational queries directly on the results page. Your page still ranks. It still earns an impression. The searcher simply never clicks, because the AI summary already gave them what they came for.
This guide is written for the moment after the panic. At CrawlCrest, an AI SEO consultancy that helps brands get found in ChatGPT, Google AI Overviews, and Perplexity, we have walked many brands through this exact diagnosis, and the honest news is mixed. Some of that traffic is gone for good. A meaningful part of it can be recovered, redirected, or replaced with visitors who convert better than the ones you lost. Let us work through it step by step.
Key Takeaways
- If your traffic dropped after AI Overviews rolled out, the most likely cause is that Google now answers informational queries on the results page, so impressions hold steady while clicks fall.
- A Pew Research Center study found users click a traditional result on only 8 percent of searches that show an AI summary, compared with 15 percent of searches without one.
- Ahrefs measured a 34.5 percent lower clickthrough rate for top-ranking pages when an AI Overview is present, based on an analysis of 300,000 keywords.
- You can confirm the cause in about 15 minutes inside Google Search Console by comparing impression and click trends, then isolating informational queries with collapsing CTR.
- Top-of-funnel informational pages get hit hardest, while branded, transactional, and local queries are largely insulated from AI Overview losses.
- The traffic is not entirely lost. A brand mention inside an AI answer is the new impression, and visitors who arrive after seeing you cited in an AI answer tend to be further along the buying journey.
- The recovery play is to get cited inside AI Overviews using answer capsules, schema, fact density, and third-party corroboration, not to fight the feature.
- Long-term resilience means rebuilding a traffic mix across AI engines, Reddit, and brand demand instead of depending on one Google surface.
Why did your traffic drop after AI Overviews?
Your traffic dropped because AI Overviews resolve the searcher's question directly on the results page, so your page keeps its ranking and keeps earning impressions while fewer people click through to read it. The query gets answered above your listing, and for a large share of informational searches, the AI summary is good enough that the visit never happens.
The independent data on this is unusually consistent. A Pew Research Center analysis of real browsing behavior found that users clicked a traditional search result on just 8 percent of visits to results pages with an AI summary, versus 15 percent on pages without one. Even links inside the summary itself were clicked on only 1 percent of visits. Users were also more likely to end their browsing session entirely after seeing an AI summary, which means many of those searches now terminate inside Google.
SEOs have a name for the resulting Search Console pattern. It is called the Great Decoupling, and it describes impressions and clicks splitting apart after rising and falling together for two decades. Your content can appear both in the AI Overview's citation list and in the organic results, which inflates impressions, while the answer on the page suppresses the click that used to follow.
The part worth sitting with is that nothing you did caused this. Your rankings did not collapse. Your content did not get penalized. The surface you built your acquisition on changed shape underneath you, and the fix starts with proving exactly where the damage is.
How do you confirm AI Overviews caused the drop?
You confirm it with a 15-minute diagnosis in Google Search Console, because traffic drops have many causes and you should never treat the AI Overview explanation as automatic. A penalty, a botched migration, lost rankings, or seasonality can produce similar topline numbers, and each needs a different fix.
Run this sequence before you change anything on the site.
- Open the Performance report, set the date range to the last 12 months, and look at impressions versus clicks. The AI Overview signature is impressions flat or rising while clicks decline. If impressions fell alongside clicks, you have a rankings problem, not an AI Overview problem.
- Check average position for your money queries. If positions are stable while CTR collapses, the click is being absorbed on the results page rather than lost to a competitor.
- Filter queries by informational patterns such as what, how, why, best, and guide. Compare their CTR over the last 6 months against the prior period. AI Overview damage concentrates almost entirely in these question-style queries.
- Now run the same comparison on branded queries. If branded CTR is steady while informational CTR fell, the diagnosis is essentially confirmed.
- Finally, search 5 to 10 of your highest-impression informational queries in an incognito window and note which now trigger an AI Overview, whether you are cited inside it, and who is.
By the end of this exercise you will know three things. Which queries trigger AI Overviews, how much CTR each one lost, and whether you are currently being cited or being replaced. That last distinction decides your entire recovery strategy.
If this is happening on your website, book a free audit with CrawlCrest and we will map exactly which queries are leaking clicks and where your recovery upside is.
Which pages get hit hardest?
Informational top-of-funnel pages absorb almost all of the damage. The Ahrefs study that measured a 34.5 percent CTR reduction for top-ranking pages also found that 99.2 percent of keywords that trigger AI Overviews are informational in intent. That tells you precisely where to focus and, just as importantly, where not to panic.
In practice, the hardest-hit page types look like this.
- Definition and explainer posts. What is X, how does X work, X meaning. These are the easiest queries for an AI summary to answer completely, so the click disappears most often.
- Generic how-to guides. Step-by-step content where the steps fit comfortably inside an AI Overview gets summarized away, especially when the steps are not tied to a specific tool or dataset.
- Stat roundups and glossaries. Pages that exist to deliver one fact get that fact extracted and displayed, with the citation earning visibility but few clicks.
- Thin listicles. Best-of lists with shallow descriptions are easy to compress into a single AI answer.
Now the insulated side, because it matters for morale and for strategy. Branded queries, where someone searches your name, still deliver the click because the searcher wants you specifically. Transactional and high-stakes commercial queries trigger AI Overviews far less often. Tools, calculators, templates, original research, and product pages keep their traffic because a summary cannot substitute for them. Your recovery plan should defend and expand exactly these categories while you fight for citations on the informational side.
Is the traffic actually lost or has the funnel moved?
A portion is genuinely gone, but a larger portion than most site owners realize has moved rather than vanished, and it moved into the AI answer itself. When an AI Overview cites your brand, that citation works like the new impression. The searcher reads your name attached to the answer, builds familiarity, and often comes back later through a branded search or a direct visit. The click you lost at the top of the funnel frequently reappears further down it, attributed to a different channel.
This shows up in client data. When CrawlCrest ran a two-month AI visibility sprint for Monk, an AR automation SaaS, the brand started showing up inside AI Overviews and ChatGPT answers for its category queries, and branded clicks grew 60 to 70 percent. The full numbers are in the Monk case study. The informational click did not return in its old form. It came back as brand demand, which is a better click anyway.
The quality argument matters here. A visitor who arrives after seeing your brand recommended inside an AI answer has already absorbed context, comparison, and a soft endorsement before touching your site. In our client work, these visitors behave like referrals rather than cold search traffic, moving to demo and contact pages faster. Raw sessions were always a vanity layer over what you actually want, which is qualified buyers. We covered the same trap from the other direction in our post on traffic but no leads, where high traffic concealed a broken funnel. AI Overviews force the healthier question. Not how many clicks did we get, but how much demand did we create.
How do you get cited inside AI Overviews instead of losing to them?
You get cited by making your pages the easiest, most trustworthy source for an AI system to extract and attribute. Google assembles AI Overviews from pages it can parse cleanly and verify externally, so citation optimization comes down to four workstreams. For a step-by-step walkthrough, our guide on how to show up in AI Overviews covers the full process.
How do you write answer capsules?
Open every informational section with a direct 2 to 3 sentence answer to the question in the heading, then expand below it. This capsule structure mirrors how AI systems lift content. Pages that bury the answer in paragraph six rarely get quoted. Audit your top 20 informational pages and rewrite the opening of each section so a machine could quote it verbatim and a reader would still nod along.
Why does schema markup matter?
Structured data tells Google what your content claims to be. Article, FAQPage, HowTo, Organization, and Person schema give the systems assembling AI answers clean entity and authorship signals. Schema does not guarantee a citation, but it removes ambiguity, and ambiguity is what gets you skipped.
What is fact density?
AI systems prefer sources that contain specific, verifiable information. Numbers, dates, named entities, original data, and firsthand observations. A page with your own survey results, real client outcomes, or concrete benchmarks gives an AI Overview something to cite that a generic rewrite cannot supply. Thin content does not just rank poorly now. It is invisible to the extraction layer entirely.
Why does third-party corroboration decide citations?
AI systems cross-reference. A claim that exists only on your site is a claim. A claim echoed on Reddit, in industry publications, in reviews, and on partner sites is a fact the model can safely repeat with your name attached. This is why mentions now matter as much as links. Our guide on getting mentioned in ChatGPT walks through building that corroboration footprint, and the same footprint feeds Google AI Overviews.
How do you rebuild a traffic mix that does not depend on one Google surface?
You rebuild it by treating Google organic as one channel among several rather than the foundation everything sits on, because the deepest lesson of this traffic drop is concentration risk. Any acquisition strategy that depends on a single surface is one product decision away from a crisis.
Three channels deserve immediate investment.
- AI engines as a distribution channel. ChatGPT, Perplexity, Gemini, and Copilot answer millions of commercial questions daily, and they recommend brands by name. Optimizing for these engines means entity clarity, consistent positioning across the web, and citable proof. The brands that show up in AI answers today are compounding an advantage while competitors keep refreshing rankings dashboards.
- Reddit and community presence. AI systems lean heavily on community discussion as corroboration, and Reddit threads rank prominently in Google itself. Genuine participation where your buyers ask questions builds the third-party footprint that both humans and models trust. It cannot be faked at scale, which is exactly why it works.
- Brand demand as the moat. Branded search survives every algorithm shift because the searcher has already chosen you. Everything that creates brand searches, from original research to podcasts to a memorable point of view, is now defensible SEO. Measure branded query growth as a primary KPI, not a footnote.
None of this means abandoning Google. It means your next quarter of effort should build assets that work across surfaces, so the next results-page redesign is an inconvenience instead of an emergency.
What should you stop doing right now?
Stop publishing thin informational posts into zero-click queries, because you are manufacturing inventory for a store with no checkout. Every generic what-is and how-to post aimed at a query that already triggers an AI Overview costs real money to produce and returns an impression with no visit. Audit your content calendar this week and cut anything whose target query is already answered above the fold.
Stop reporting impressions as a success metric while clicks fall. The decoupling makes impression growth close to meaningless on informational queries, and celebrating it hides the problem from decision makers who control budget.
Stop blaming the team for a platform shift. The instinct after traffic drops is to rewrite titles, churn out more posts, or fire the provider. If the diagnosis shows stable rankings and collapsing informational CTR, more of the old playbook produces more of the old result. Redirect that energy into citation optimization and channel diversification, where effort still compounds.
And stop waiting for it to revert. AI Overviews have expanded steadily since launch, and search behavior is migrating toward answer engines across every platform. The brands treating this as the new baseline are already pulling ahead of the ones waiting for 2023 to come back.
How does CrawlCrest help you recover traffic lost to AI Overviews?
CrawlCrest is an AI SEO consultancy built for exactly this problem. We help brands get found in ChatGPT, Google AI Overviews, and Perplexity, and recover the demand that traditional rankings reports can no longer see or explain.
The engagement starts with an audit. We map your Search Console data to identify precisely which queries lost clicks to AI Overviews, separate real losses from funnel shifts, and benchmark how visible your brand currently is across AI engines. From there our AI Overview optimization work rebuilds your most valuable pages around answer capsules, schema, and fact density so they earn citations instead of losing clicks, and our AI SEO consulting builds the third-party corroboration footprint across Reddit, communities, and industry publications that AI systems require before they will recommend a brand by name.
The results are measurable. After our two-month sprint, Monk appeared inside AI Overviews and ChatGPT answers for its core category queries and grew branded clicks 60 to 70 percent. That is what recovery looks like in 2026. Not clawing back every informational click, but converting AI visibility into brand demand that compounds.
If your traffic dropped after AI Overviews and you want a clear picture of what is recoverable, get a free audit and we will show you exactly where you stand, which queries to fight for, and which channels will replace what Google absorbed.
Final thoughts on traffic dropped after AI Overviews
Watching years of traffic evaporate while every dashboard insists nothing is wrong is genuinely disorienting, and most advice either tells you to panic or to ignore it. The truth sits in the middle. The informational click is shrinking and will not fully return, but the demand behind it still exists. It now flows through AI answers, brand searches, and communities, and it rewards the brands that show up inside those surfaces with visitors who convert better than the ones the old funnel delivered.
Run the 15-minute diagnosis. Defend the page types that still earn clicks. Restructure your best content to get cited, build corroboration the models can verify, and grow channels Google cannot take away. The phrase traffic dropped after AI Overviews describes the symptom, but concentration risk was always the disease, and curing it leaves you stronger than before the drop.
When you are ready to turn the diagnosis into a recovery plan, talk to CrawlCrest and we will build it with you.
