AI Overviews killed my traffic. Here's why it wasn't a penalty.
Australian SMBs are losing organic traffic at an unprecedented rate, and almost none of it is a Google penalty. It's something harder to fix.
Three of our clients lost more than 40% of their Google traffic in the last quarter. None of them did anything wrong. None of them got a manual action. None of them broke any guideline. Their traffic just left — and it's not coming back.
If you've been searching "why did my organic traffic drop" and finding only generic checklists about meta tags and broken links, this is for you. The thing that's happening to you is structural, not technical. And the sooner you stop hunting for a penalty that doesn't exist, the sooner you can do something about it.
Why your traffic left (and why it isn't coming back)
Google now answers a large share of informational questions directly inside the search results page, using AI Overviews. The user reads the AI summary, gets their answer, and never clicks through to the underlying website — even when that website is the source the AI quoted.
This isn't a penalty. There's no manual action in Search Console. Your rankings are often unchanged. You're still on page one. But the click-through rate has collapsed, because the user no longer needs to click.
Across the AU SMB sites we manage, click-through rates for informational queries dropped between 30% and 60% over the 12 months following AI Overviews rollout, even for pages whose ranking positions didn't move. The same pattern shows up in third-party publisher reports across the AU and US markets, including reports from major news publishers who have been forced to restructure entire content strategies in response.
Three things you can stop doing right now
These are wasted effort:
- Hunting for a penalty. There isn't one. Stop refreshing Search Console looking for a manual action that won't appear. Google explicitly states that manual actions are listed in the dedicated Manual Actions report — if it's empty, you don't have one.
- Adding more meta keywords or rewriting title tags. The traffic loss is downstream of impressions, not rankings. Rewriting tags doesn't get clicks back when the user already has their answer.
- Pumping out more "how-to" blog posts to chase informational queries. AI Overviews are eating that entire category. You'll write more and earn less.
The three moves that actually work
These are the patterns that recovered traffic, or at least stabilised it, across our client base in the six months following the rollout.
1. Rewrite for high-intent queries, not informational ones
AI Overviews are voracious for "what is X" and "how do I Y" queries. They're much weaker on "best X for Y in [city]," "X near me," and comparison queries. Reallocate your content effort toward those formats. They still drive clicks because the user has a buying decision, not a question.
For an AU plumber, the difference is between writing "how to fix a leaking tap" (an informational query that AI Overviews now answers in three sentences) and writing "best emergency plumbers in inner Melbourne after-hours" (a comparison query that AI summaries struggle with because the answer is contextual and personal). The second one still earns clicks; the first one is fighting yesterday's war.
2. Become the source AI Overviews quote
If you can't beat AI Overviews, get inside them. Original data, named experts, primary research, and clearly-cited statistics are disproportionately likely to be cited in the AI summary. A blog post with a stat the AI can't get elsewhere is more valuable now than ten posts that summarise existing knowledge. Run a small survey, audit your customer data, publish a quarterly index — these are the artefacts AI engines cite.
3. Build owned audience channels
Email lists. Direct subscribers. RSS followers. LinkedIn followings. Anything that lets you reach an audience without Google's permission. The lesson of the AI search shift is the same lesson publishers learned five years too late from Facebook: a borrowed audience is a vulnerable one.
In a 2020 IEEE Access literature review I co-authored with Nina Evans, cited 260+ times, we found that the integration of AI and big data in marketing isn't a tool layer that sits on top of existing channels — it eventually becomes the channel itself. What we're seeing now with AI Overviews is the operationalisation of that pattern in search specifically. The substrate is no longer "a list of websites"; the substrate is "an answer." Sites that adapt to producing answer-quality content, rather than article-quality content, will recover. The rest will keep losing impressions, slowly, indefinitely.
What this means for Australian businesses
In the AU market, where many SMBs run on tight marketing budgets and have one or two channels driving most leads, this shift is more dangerous than in larger markets. There isn't a paid-search budget waiting in reserve. There isn't a 12-person marketing team to pivot.
If you run a service business in Melbourne, Sydney, Brisbane, or Perth, the immediate priority is auditing your current organic traffic by query intent. Informational queries are bleeding. Local-intent queries are largely intact. Comparison queries are growing. Your content mix should reflect that — and most AU SMB sites we audit haven't even looked at the breakdown.
The Search Console "Performance" report lets you slice by query. Group your queries into three buckets — informational, comparison, transactional/local — and check the CTR trend on each over the last 12 months. The picture is usually stark.
How to do the audit yourself in 30 minutes
If you want to run the diagnostic before talking to an agency:
- Open Search Console and go to Performance → Search results.
- Set the date range to the last 12 months. Note the average CTR.
- Switch to "Compare" mode. Compare the most recent 3 months against the same 3 months a year ago.
- Filter by query containing "how" — your CTR has likely dropped sharply.
- Filter by query containing "best" or "vs" — these have likely held up or grown.
- Filter by queries containing your suburb or city — usually intact.
If your informational-query CTR has dropped more than 25% in that comparison while ranking positions are roughly the same, you have an AI Overviews problem, not a penalty. The recovery path is content-mix reallocation, not technical SEO repair.
Want this audit run on your site?
Catnap is auditing AU SMB sites for AI search visibility — informational query CTR decay, AI Overview citation rate, and the content mix that's still earning clicks. Three-page report, six-criteria framework, prioritised fix list.
The bottom line
If your traffic dropped and there's no manual action, you don't have a penalty — you have a structural shift to absorb. Stop wasting weeks looking for a fix that doesn't exist. Move your content effort toward queries AI Overviews can't satisfy, become the source those AI summaries cite, and start building an audience you can reach without Google's permission. The agencies that solve this are the ones you should be talking to. The ones still fixing meta tags are working on a problem that ended in 2024.
About the author
Dr Andrej Miklosik is the founder of Catnapweb and a Professor at the Department of Marketing and Commerce, Faculty of Management, Comenius University in Bratislava. His peer-reviewed work on machine learning in marketing has been cited 1,628 times, with an h-index of 17 across 34+ published papers in IEEE Access, Heliyon, and other journals.
Read more about Andrej →Sources
- Google. (2024). Generative AI in Search: Let Google do the searching for you. Google Blog.
- Google Search Central. AI features in Google Search documentation.
- Google Search Central. Manual Actions report documentation.
- Miklosik, A., & Evans, N. (2020). Impact of big data and machine learning on digital transformation in marketing: A literature review. IEEE Access, 8, 101284–101292.
- The Verge. (2024). Google's AI Overview is killing publisher traffic.
- Google Search Central. Get on Google Discover.
