A year on from the broad rollout of AI Overviews in Australia, we sat down and ran the numbers across our client book. 240 commercial queries we have tracked since 2023. Before and after AIO. The picture is more nuanced than the LinkedIn doom posts suggest.
The headline numbers
- Average click through rate on positions 1 to 3 fell by 18% on queries that triggered an AI Overview.
- Average CTR on positions 4 to 10 fell by 31% on those same queries.
- Branded queries were essentially unaffected. The fall was concentrated in informational and early stage commercial queries.
- Pages cited inside the AI Overview itself recovered roughly 60% of the lost CTR. Sometimes more.
So the doom posts are partly right. There is real loss on the queries that trigger an AIO. But the loss is unevenly distributed, and being inside the AIO is now its own SEO objective with its own playbook.
What survived
Three categories of query barely moved.
First, anything with strong commercial intent. “Buy”, “near me”, “best [product]” in 2026, “[brand] reviews”. Google still wants to show product results and local pack results for these because the user is buying something. The AIO either does not appear or sits politely above a healthy SERP.
Second, anything navigational. People still type brand names into Google. AIO does not really compete here.
Third, anything where the answer requires real time data. Sports scores, share prices, weather, opening hours. AIO defers.
What got eaten
The losses concentrated in the queries that should have been the easiest to win. Definitions. How to questions. Comparison queries. “What is X”. “How does Y work”. “X vs Y”. For most of the last decade these were the cheap entry points to a content program. Build a clean definitional page, rank in three months, ride the traffic for two years.
Those pages are now the AIO.
The cheap content plays of the last decade are the ones that died. The hard content plays are basically untouched.
What we changed
Three changes across the book of work.
1. We stopped writing definitional content as a primary play.
If a query is “what is X”, the AIO is going to win it. Writing a 1500 word definitional page in 2026 is writing a page that will rank but never get clicked. We still cover definitions, but as the lead paragraph of a deeper page, not as the page itself.
2. We chase citation in the AIO instead of (or in addition to) the blue link.
If we cannot beat the AIO, we want to be quoted inside it. The optimisation is real. Lead with the answer. Use schema. Make the quotable sentence five to twenty words. We have client pages that get cited inside Google AIOs three to five times a week and recover most of the click loss.
3. We rebuilt the keyword strategy around demonstration not definition.
The query patterns that still convert are the ones where the user wants to see something. “Show me how X looks”. “Examples of X”. “Worked example of Y”. These pages cannot be summarised by an AIO without losing the value, so the AIO either does not trigger or links out.
The honest bit
AIO is not a catastrophe. It is also not nothing. The agencies telling clients it is fine are the ones who have not done the data work. The agencies telling clients SEO is dead are selling something else.
The truth is that informational SEO got harder, commercial SEO got more competitive, and a new optimisation surface (the AIO itself) opened up that pays back if you bother with it.