Pre-AIO baseline vs post-AIO observed CTR by SERP position
Avg drop
−54%
Worst hit
Pos 4-5
Recovery
AIO citation
Live demo · the CTR collapse on informational queries from the pre-AIO baseline to the post-AIO observed numbers, by SERP position.
On informational queries where AI Overviews appears, top organic results have lost 30-50% of CTR on average, with position 4-5 hit hardest. Transactional and local queries are barely affected. The recovery playbook isn't to outrank AIO - it's to be the cited source inside it. Same content shape, different distribution surface.
1. The CTR damage by position
On informational queries where AI Overviews appears, the top organic result has lost roughly 30-50% of click-through rate compared to the same position pre-AIO. Position 2-5 has seen even sharper drops (40-60%) because the AIO card pushes them below the fold. Position 6+ is essentially decimated for these query types. The damage is concentrated on the queries that used to drive informational content traffic.
2. Which queries lost CTR
Definitional queries ("what is X"), how-to queries, comparison queries ("X vs Y") and quick-answer queries are the worst hit. These are exactly the queries where users used to scan organic snippets to grab a quick answer - AIO now does that work directly inside the SERP, so the click is gone. Transactional queries ("buy X", "X price"), local queries ("X near me") and long-tail conversion queries are barely touched. The split maps roughly to the broader behaviour shift covered in our AI search vs traditional search guide.
3. The recovery playbook
You don't recover by trying to outrank AIO - that ship has sailed. You recover by being the cited source inside the AIO card. AI Overviews loves concise, declarative, schema-rich content with clear answers to question-shaped queries. The format playbook is identical to the one in our write content AI assistants cite guide - lead with the answer, atomic facts, structured comparisons. Long intros and dwell-time-optimised SEO prose get filtered out almost completely.
4. Schema is the AIO accelerant
FAQPage and HowTo schema get pulled into AIO answers disproportionately often because they're machine-readable representations of exactly the question-and-answer shape AIO wants to surface. Pair them with a clean Organization schema and you ship the parsing layer that AIO needs to confidently cite you. The full pattern lives in our schema markup for AI search guide.
5. Should you opt out of AIO?
Technically you can - the Google-Extended user agent directive in robots.txt opts you out of Gemini training, and noindex works on specific pages. The trade-off is brutal though: blocking AIO usually removes you from Gemini app results and other AI search surfaces using the same crawl pipeline. For most brands the right play is to optimise for AIO citation instead of opting out, and to win the traffic back as a cited source. The full strategy for showing up inside AIO is in our rank in AI Overviews guide.
Recap
AI Overviews has cost informational queries 30-50% of organic CTR on average, with position 4-5 hit hardest. Transactional and local queries are largely unaffected. The recovery is not to outrank AIO but to be the cited source inside it - declarative content, FAQ + HowTo schema, the same playbook that wins in ChatGPT, Claude and Perplexity. Brands optimising for AIO citation right now are the brands clawing the traffic back; brands waiting for the SERP to revert are losing the channel for good.
Get cited inside AI Overviews
Geolify GEO packages ship the schema, content overhaul and entity build that turn your top informational pages into the cited source inside Google AI Overviews. From $499.
FAQ
How much has AI Overviews actually hurt organic CTR?
On informational queries where AI Overviews appears, the top organic result has lost roughly 30-50% of its click-through rate compared to the same position pre-AIO. Position 2-5 has seen even sharper drops (40-60%) because the AIO card pushes them below the fold. Transactional and local queries are barely affected because AIO doesn't show up reliably for them. The damage is concentrated on the queries that used to drive informational content traffic.
Can I block my content from AI Overviews?
Yes, technically - you can use the Google-Extended user agent directive in robots.txt to opt out of Gemini training, and you can use noindex on specific pages. But the trade-off is brutal: blocking AI Overviews also tends to remove you from Gemini app results and most other AI search surfaces that rely on the same crawl pipeline. For most brands, the better play is to optimise to be the cited source inside AIO rather than to opt out entirely.
What kind of content gets cited inside AI Overviews?
Concise, declarative, schema-rich content with clear answers to question-shaped queries. AI Overviews loves FAQ schema, HowTo schema, and pages where the first 100 words contain the citable answer. The format playbook is identical to the one we cover in our 'write content AI assistants cite' guide - lead with the answer, atomic facts, structured comparisons. Long intros and dwell-time-optimised SEO prose get filtered out almost completely.
Should I still optimise for organic ranking if AIO is taking the clicks?
Yes - but the goal shifts. Position 1 organic is no longer the prize on its own; being inside the AI Overview card is. The good news is that the same content that ranks well organically tends to be exactly what AIO cites, because Google uses very similar quality signals for both. Treat organic ranking as the necessary precondition for AIO inclusion, then specifically optimise for the AIO citation format on top.
Which queries lose the most CTR to AI Overviews?
Definitional queries ('what is X'), how-to queries ('how do I X'), comparison queries ('X vs Y'), and quick-answer queries ('best X for Y'). These are exactly the queries where users used to scan organic snippets to grab a quick answer - AIO now does that work directly inside the SERP, so the click is gone. Transactional queries ('buy X', 'X price'), local queries ('X near me'), and long-tail conversion queries are barely touched.