Why Google's AI Overviews Are Sending You to the Wrong Audience
AI Overviews aren't killing your traffic evenly. They're stealing the browsers and leaving the buyers, but only for sites cited inside the answer box.
Built BakingSubs to 162,500 Copilot citations and accelerating. Now teaching the system behind it.
- ai-visibility
- google-ai-overviews
- strategy
- client-generation
Most coaches looking at their Google Analytics right now think AI Overviews are eating all their traffic. They're not. They're eating the wrong half of it, and the half they leave behind is the half that was actually going to hire you.
Key takeaways
- AI Overviews mostly siphon top-of-funnel informational searches (definitions, "what is", "how does"), and those searchers were rarely buyers anyway.
- Buyer-intent searches (comparisons, "best for", "near me", "who should I hire") still produce clicks, but mostly to the sites cited inside the Overview itself.
- If your site isn't named in the Overview for a buyer query, you're effectively invisible for that query, even if you rank #3 organically.
- The fix isn't writing more SEO content. It's writing the specific kind of content AI engines lift directly into answers.
- BakingSubs earned 144,321 Microsoft Copilot citations in a quarter using The Citation Cluster Method, the same approach that gets coaching sites cited inside Google's Overviews.
What AI Overviews actually took from you
Google's AI Overviews mostly took the search traffic that was never going to convert. Informational queries like "what is somatic coaching" or "how does executive coaching work" used to land on your blog, bounce in 40 seconds, and never come back. Now they get answered in the Overview box and the searcher moves on. Your blog post didn't lose a buyer. It lost a researcher.
Look at your traffic drop by query type, not by total volume. The queries that fell off a cliff are almost all "what", "how", "why", and "define" patterns. The queries that held up (or are growing) are the ones with buyer signals: "best", "vs", "near me", "for [specific situation]", "review of", "is [thing] worth it". That's the half that matters.
Consider Priya, a life coach in Toronto working with second-generation South Asian women in finance careers. Her blog post titled "What Is Career Coaching?" used to bring her 800 monthly visits. After AI Overviews rolled out, it brought her 90. She panicked. Then she looked at her booked discovery calls and they hadn't moved. None of those 800 visitors were ever calling her. The 12 a month who searched "career coach for South Asian women Toronto" still were, and that post had grown.
Why buyers still click, and where they click to
When a buyer types a high-intent query into Google now, the Overview appears, and inside it Google names 2 to 4 specific sources. Buyers click those named sources at a high rate because the Overview just pre-vetted them. Everything below the Overview is essentially noise to a buyer who's already been handed a shortlist.
So the question stopped being "do I rank on page one" and started being "am I one of the 2 to 4 sites Google decided to cite". A site at organic position #2 that isn't cited in the Overview will lose clicks to a site at position #11 that is. This is the part most SEO advice still hasn't caught up to.
The Overview pulls its cited sources based on a different set of signals than classical ranking. It wants sources that answer the specific question directly, in clear language, with author signals it can verify, and with structured page elements it can parse. A long, meandering, story-heavy post that ranks well on classical SEO can get completely passed over for citation in favor of a tighter post lower in the rankings.
The kind of content AI Overviews actually cite
AI Overviews lift content that reads like an answer, written by an identifiable expert, on a page that's clearly about one specific thing. That's it. Most coaching content fails on at least two of those three.
Here's what each piece looks like in practice:
| Signal | What Overviews want | What most coaches publish |
|---|---|---|
| Direct answer | First 1 to 2 sentences after the heading answer the question | Story-driven opener, answer buried in paragraph 4 |
| Author signal | Named human author with verifiable credentials and a real bio | "Posted by admin" or a brand-only byline |
| Topic focus | Page is about one specific question for one specific audience | Catch-all post trying to rank for 5 different queries |
James, a workplace mediator in Manchester, rewrote 6 of his existing posts to follow that pattern. He didn't add new content. He restructured what was there. Within 9 weeks, 4 of those posts were being cited inside Google's Overviews for queries like "how to handle a co-founder dispute" and "when to bring in a workplace mediator". His total organic sessions went down. His booked consultations went up. That's the trade you want.
If you've been told that the answer is more content, that's the old playbook. The new one is fewer, sharper pages that look more like reference entries than blog posts. The full mechanic is what I cover in the Citation Cluster Method, which is how BakingSubs ended up cited 144,321 times by Microsoft Copilot in a single quarter without a single backlink campaign.
How to tell which of your pages are being cited (and which aren't)
Google doesn't show you Overview citation data in Search Console. You have to check by hand. Pick your top 20 buyer-intent queries, type each one into Google while signed out, and note which sources Google cites inside the Overview box. Then cross-reference that list against your own URLs.
You'll usually find one of three patterns:
- You rank well but aren't cited. The page exists but isn't structured for extraction. Fixable.
- You don't rank and aren't cited. The page either doesn't exist or is too thin. Build it.
- You're cited. Leave it alone and study what made that page different from the others.
The third pattern is the most useful one because it tells you what your own voice and structure look like when they work. Most coaches have at least one page that gets cited and they don't know why. Find it. Copy the pattern across your other buyer-intent queries.
This is also what the free AI Visibility Check automates. It runs the buyer-intent questions for your niche across ChatGPT, Claude, Perplexity, and Copilot and tells you which engines are citing you, which are citing your competitors, and which are returning nothing useful at all. The same gaps usually show up in Google's Overviews.
Why this hits expert-led businesses harder than big brands
Big brands get cited by Overviews on reputation alone. Google's models already know who they are. Solo coaches, consultants, and small expert-led businesses don't have that head start, so the citation comes down to on-page signals: how clearly the page answers the question, who the named expert behind it is, and how focused the surrounding cluster of pages is on the same topic.
The good news is that those signals are entirely within your control. The bad news is that nobody else's content advice is currently telling you to optimize for them, because most SEO writers haven't watched a single client's traffic drop the way yours has and still tried to figure out what actually moved the buyer needle.
One contrarian take I'll say plainly: stop trying to recover the informational traffic you lost. It wasn't paying you. Spend that same energy making sure the 30 to 50 buyer-intent queries in your niche have a page on your site that's built to be cited. That's a much smaller, more achievable project than rebuilding your top-of-funnel volume, and it directly affects your pipeline.
What to do this week
Pick your top 10 buyer-intent queries. For each, do the manual Overview check. Group your pages into the three patterns above. Then rewrite one page from pattern #1 (ranks but isn't cited) using the direct-answer-first structure. Watch what happens over the next 4 to 6 weeks.
If you want a faster diagnosis, the AI Visibility Check covers 8 discovery-intent questions per engine and tells you exactly where you stand on ChatGPT, Claude, Perplexity, and Copilot. The patterns it surfaces map closely to what Google's Overviews are doing, because all five engines weigh similar signals when deciding who to cite.
Frequently asked questions
Are Google AI Overviews killing traffic for every type of page?
No. Overviews mostly absorb informational queries (definitions, explainers, "how does X work") where the searcher just wanted a quick answer. Buyer-intent queries like "best [type] coach for [situation]" still drive clicks, but mostly to the 2 to 4 sources Google names inside the Overview. If your traffic drop is concentrated on informational pages, you probably lost mostly non-buyers.
How do I get my site cited inside a Google AI Overview?
Three things matter most: answer the specific question directly in the first 1 to 2 sentences after the heading, attach a named human author with verifiable credentials, and make sure the page is focused on one specific question rather than trying to cover five. The same signals that work for getting found on ChatGPT as an expert work for Google's Overviews.
Does ranking #1 in regular search still matter if I'm not in the Overview?
It matters less than it used to for buyer-intent queries. When an Overview appears, the cited sources inside it capture most of the click intent, and the classical organic results below get a fraction of what they used to. A site cited in the Overview but ranking at position #11 will often outperform an uncited site at position #2. The goal is to be inside the box, not below it.
Should I delete my old informational blog posts that lost traffic?
Usually no. They still answer real questions and they often support your topical authority, which feeds your buyer-intent pages. What you should do is stop investing more time in them and shift that effort toward building topical clusters AI engines actually cite around your buyer queries.
How long does it take to start getting cited after I fix a page?
Anywhere from a few weeks to a few months. Google's Overview tends to update faster than its classical ranking algorithm, so a well-structured rewrite of an already-indexed page can get cited within 4 to 8 weeks. New pages on a site with no prior authority take longer because the engine has to first decide the site is a credible source.
The biggest mistake right now is reacting to total traffic loss instead of reading what kind of traffic you actually lost. Pull your top 20 buyer-intent queries, check the Overviews by hand, and find the gap between where you rank and where you're cited. Fix one page this week using the direct-answer structure and watch the citations over the next month. If you want the full diagnosis automated across the four major AI engines, that's what the AI Visibility Check is built for.