- Pew Research Center found that users clicked a traditional Google result in 8% of visits when an AI summary appeared, compared with 15% when no AI summary was shown.
- Seer Interactive’s data suggests the bigger issue is no longer just ranking. Marketers now need to know which queries trigger AI Overviews, whether their brand is cited, and where clicks are falling despite stable positions.
Click-through rates dropped from 15% to 8% in one year once AI Overviews enter the results. Your rankings didn’t tank. The game shifted around them. The brands winning right now aren’t the ones with the most backlinks. They’re the ones AI trusts enough to quote. Structured data.… pic.twitter.com/JiVq2uTQeG
— Neil Patel (@neilpatel) July 4, 2026
A July 4 post from Neil Patel captured a problem many SEO teams are now seeing in their own dashboards: rankings can look stable while clicks quietly fall away.
The post pointed to click-through rates dropping from 15% to 8% when AI summaries appear in Google results. That figure lines up with Pew Research Center’s analysis of U.S. Google searches, which found that users clicked a traditional result in 8% of visits when an AI summary appeared, compared with 15% when no AI summary was present.
The easy version of the story is that AI Overviews reduce clicks. Most marketers know that by now. The more useful question is where the clicks are disappearing, which pages are still exposed to them, and which competitors are being used as sources instead.
The Problem Is Not Always the Ranking
For years, the SEO diagnosis was fairly simple. If clicks dropped, teams checked rankings. If rankings were stable, they looked at seasonality, demand, snippets, ads, or tracking issues.
AI Overviews add another failure point. A page can still rank in the same average position, but the result may now sit below a generated answer that resolves the user’s query before the organic listing gets a chance.
That is why position tracking alone is becoming less useful for informational searches. A page in position three on a query with no AI Overview is not in the same situation as a page in position three below a generated answer, source cards, video blocks, ads, People Also Ask, and Reddit results.
The ranking number may be similar. The click opportunity is not.
What the CTR Data Actually Shows
Seer Interactive analyzed 3,119 informational and educational queries across 42 organizations from June 2024 through September 2025. The dataset covered 25.1 million organic impressions and 1.1 million paid impressions.
For queries where AI Overviews appeared, organic CTR fell from 1.76% to 0.61%. Paid CTR dropped from 19.70% to 6.34%.
That confirms the obvious part: AI Overviews make clicks harder to earn.
But the more interesting detail is what marketers should do with that information. The useful audit is not “Do AI Overviews hurt traffic?” It is:
- Which of our high-impression queries now trigger AI Overviews?
- Which pages still rank but lost CTR?
- Are we cited inside the AI Overview?
- If not, which domains are being cited instead?
- Are those domains competitors, publishers, forums, review sites, or documentation pages?
That turns the issue from a general AI-search complaint into a page-level diagnosis.
Citation Helps, but It Is Not a Magic Ranking Factor
Patel’s post argues that the winning brands are the ones AI systems trust enough to quote. That is directionally right, but it needs a careful reading.
Seer found that brands cited in AI Overviews saw 35% higher organic CTR and 91% higher paid CTR than non-cited brands on AI Overview queries. That is a useful signal. It does not prove that being cited directly caused the lift.
The brands being cited may already be stronger. They may have better-known names, better content, stronger topical authority, cleaner pages, more useful data, or higher trust before the AI Overview appears.
That distinction matters because it changes the work. The answer is not to chase “AI citations” as if they were backlinks. The answer is to make the pages that already rank more citeable.
Top Rankings Still Feed AI Overviews
This is where the “backlinks are dead” version of the story breaks down.
Ahrefs analyzed 1.9 million citations from 1 million AI Overviews and found that 76.1% of cited pages ranked in Google’s top 10. The median cited URL ranked in position three.
That means AI Overviews are not replacing traditional SEO with a completely separate system. They are often selecting from pages that already perform well, then giving a smaller group of those pages extra visibility inside the answer.
For SEO teams, that creates three different cases:
- Ranking and cited: the best-case scenario. The page keeps organic visibility and gains presence inside the AI answer.
- Ranking but not cited: the danger zone. The page still appears in reports, but competitors may be taking the answer box.
- Not ranking but cited elsewhere: often a sign that Google prefers a third-party source, such as a review site, forum, glossary, documentation page, or data-heavy article.
That middle case is the one many dashboards miss. It looks fine until CTR is checked query by query.
How to Audit This Without Overcomplicating It
A practical AI Overview audit does not need to start with an expensive platform. A simple monthly workflow is enough for many sites.
Start with Search Console and export queries with high impressions, stable average position, and falling CTR. Filter for informational searches first, because those are the queries most exposed to AI summaries.
Then manually check the top 30 to 50 queries in Google. For each one, record:
- whether an AI Overview appears;
- whether your site is cited;
- which competitors or third-party sites are cited;
- whether your page answers the query quickly;
- whether the cited source has clearer structure, fresher data, stronger author signals, or a better definition section.
This is not glamorous work, but it is the part that shows what actually changed. Many traffic drops will not come from lost rankings. They will come from a SERP where the page still ranks but no longer controls the answer.
What Makes a Ranking Page More Citeable
Structured data can help, but it is not the whole answer. Adding schema to a vague article does not suddenly make it the best source.
The pages most likely to survive this shift usually do a few things well. They answer the main question early. They make facts easy to extract. They use clear subheadings. They include current numbers, named sources, author context, and definitions that do not require the reader to dig.
For commercial or B2B content, comparison sections can also matter. AI systems often need concise differences between tools, tactics, platforms, or categories. If the page only says why the brand is great, it is less useful as a neutral source.
For publishers, original reporting helps. Quotes, mini datasets, screenshots, examples, expert commentary, and first-hand testing give a page something that a generic explainer cannot offer.
That is the part many AI-search articles skip. “Write direct answers” is not enough. The page has to give the system a reason to prefer it over the many other pages that also answer directly.
What Marketers Should Change in Reporting
The reporting stack needs one more layer.
Rankings still matter. Search Console still matters. Traffic still matters. But for affected queries, teams should also track citation status and SERP shape.
A simple report can separate queries into four groups:
- Stable rankings, stable CTR: no immediate issue.
- Stable rankings, falling CTR, no AI Overview: check ads, SERP features, demand, title relevance, and competing formats.
- Stable rankings, falling CTR, AI Overview present, not cited: priority page for citation improvement.
- AI Overview present, cited competitor: analyze why that source was easier to use.
This is more useful than reporting an average CTR decline across the whole site. It shows where the business is actually losing opportunity.
Google’s AI search shift is already changing SEO before rankings show it, which means marketers need to stop treating traffic drops as a single diagnosis. Some pages need better titles. Some need stronger content. Some need more authority. Some are simply sitting below a result format that now answers the query first.
What Marketers Should Take From This
The useful takeaway from Patel’s post is not that AI Overviews are bad for clicks. That part is already clear.
The more important point is that SEO success is splitting into separate signals. Ranking, citation, click-through rate, and brand visibility can now move in different directions.
A page can rank and lose clicks. A brand can be cited and receive limited referral traffic. A competitor can become the named source even when another site holds a similar organic position.
That is why the next phase of SEO reporting needs to be more specific. Marketers should not only ask whether a page ranks. They should ask whether the page is being used.
The teams that answer that question query by query will see the change earlier. The ones looking only at average positions may keep wondering why the numbers look fine while the traffic does not.
