The first time I noticed Google’s AI Overviews stealing clicks from pages I’d spent weeks building, it wasn’t on a broad keyword. It was on something niche—a kinako ice cream recipe that had ranked steadily for years. Impressions held. Clicks dropped. The AI Overview had absorbed the answer and the reader never needed to visit. That moment changed how I think about content structure across every property I run.
I operate ADIELAS: eight websites, a Japanese-tea e-commerce brand, and an Amazon storefront, run with around twenty self-hosted AI agents on roughly $260 a month in software. SEO traffic is not a hobby metric for me—it funds the business. When AI Overviews started eating into that traffic, I needed a systematic response, not another round of theoretical frameworks from people who don’t have a P&L attached to their rankings.
What follows is what’s actually worked.
Last updated: June 2026
Key Takeaways
- AI Overviews pull from clearly structured content—FAQ sections, direct definitions, and answer-first paragraphs are more likely to be cited than prose buried behind a hook.
- Transactional pages (product comparisons, buying guides) are far less exposed to AI Overview cannibalization than informational content—protect the information layer first.
- Recipe, Product, and FAQPage schema are table stakes now; without them, Google’s AI has a harder time parsing and attributing your content.
- Internal linking between informational and product pages is your topical authority map—build it deliberately, not incidentally.
- Google Search Console’s “AI Overviews” appearance filter shows exactly which queries are at risk; restructure the highest-impression pages first.
What AI Overviews Actually Are (And Aren’t)
Google’s AI Overviews are AI-generated answer blocks that appear above or alongside traditional search results. They synthesize content from multiple sources to directly answer a query. Your page can still rank below the Overview—but the click that used to come to you for quick informational queries now often goes nowhere.
The distinction that matters for operators with real inventory and margin pressure: Overviews are most aggressive on informational intent. “What is genmaicha?” or “how do I store hojicha?” are high-risk. “Buy genmaicha tea” or “best genmaicha under $20” are lower-risk because transactional intent is harder for an Overview to satisfy—the user still needs to complete a purchase somewhere.
This is why a kinako ice cream recipe page bleeds clicks while a genmaicha product comparison page holds. The recipe answers a question. The comparison facilitates a transaction. Overviews are better at the former, and that gap is where your restructuring effort should focus.
How Google Decides What Gets Cited in an Overview
From testing content restructures across several of my sites, Google’s AI consistently favors content that:
- Answers the query in the first paragraph without burying the lead
- Uses clearly scannable structure—numbered steps, definition blocks, short paragraphs
- Has structured data markup that labels what the content is (Recipe, Product, FAQPage)
- Demonstrates topical depth through internal linking, not just keyword density
- Comes from a domain with established authority on the topic cluster, not a single orphaned page
That last point is the one most operators miss. A single well-optimized page from a domain with no topical coherence will lose to a thinner page from a site with twenty interlocking pieces on the same subject. AI Overview attribution is a cluster game, not a single-page optimization play. If your content pages have been built as independent units with no meaningful cross-linking, that’s the structural problem to fix first.
Practical Optimization: What to Actually Do
Lead with the direct answer
Rewrite your opening paragraphs so the core answer appears in the first two sentences. For a guide to storing Japanese tea, the first sentence should be: “Store loose-leaf Japanese tea in an airtight, opaque container away from light, heat, and moisture.” Explanatory depth comes after. Google’s AI needs to find the answer fast to cite it.
This felt counterintuitive when I first applied it—I’d been trained to write hooks that built to the answer. For AI Overview optimization, the hook and the answer need to be the same thing.
Add a FAQ block to every informational page
FAQ sections are high-leverage because they pre-format content as question-answer pairs, which is exactly how AI Overviews are structured. Adding FAQPage schema to existing posts and restructuring their bottom sections into explicit Q&A blocks made a visible difference in which of my pages started appearing as cited sources in Overviews on related queries.
Keep answers short—two to four sentences. If your answer to “does genmaicha go bad?” is a 200-word essay, Google’s AI will either truncate it awkwardly or skip it for a cleaner source. Say the essential thing and stop.
Implement structured data—actually implement it
Recipe schema for recipe pages, Product schema for product pages, FAQPage schema for FAQ sections. Compliance on this among established content operations is still surprisingly low. I’ve audited pages where the recipe card plugin was installed but the schema was misconfigured, product pages had no structured data at all, and FAQ blocks were plain HTML with no markup.
Validate every page with Google’s Rich Results Test. Recipe schema needs ingredients, instructions, and at minimum prep or cook time. Product schema needs price and availability. Half-implemented schema is often worse than none—it signals technical sloppiness to crawlers. Fix it completely or skip it.
The Cluster Strategy That Changed How I Build Content
The single biggest lever for AI Overview visibility isn’t page-level optimization—it’s building a genuine content cluster around a topic so that Google’s AI sees your domain as the authoritative source when it synthesizes an answer.
For a Japanese food and tea operation, that looks like:
- A cornerstone “what is genmaicha” explainer covering origin, flavor, and use cases
- A buying guide comparing genmaicha options by grade and price point
- Recipes that use genmaicha—rice, latte, cold brew
- A storage guide linking back to the product and the cornerstone
- Internal links connecting all of these to each other and to the product page
When these pages interlock with natural anchor text and consistent topic coverage, Google’s AI sees one coherent authority on genmaicha rather than five unrelated pages that happen to mention it. A matcha vs. genmaicha comparison does double duty: it answers a real query and deepens the cluster by linking to both the product page and the storage guide.
Internal links aren’t navigation. They’re your topical authority map. Build them with intent.
Measuring What’s Working
Google Search Console is the right tool here. Filter your queries by those that trigger AI Overviews using the “AI Overviews” appearance type under Search Appearance. Compare impressions to clicks on those queries. If impressions are flat or growing but clicks are declining, the Overview is absorbing the traffic.
That’s not necessarily a crisis—it’s a signal to either restructure the page to get cited inside the Overview, or pivot that page toward commercial content where Overviews are less aggressive.
After restructuring the kinako ice cream page with a direct opening answer, FAQ schema, and tighter paragraph structure, the click pattern shifted—not because the page ranked higher, but because it started appearing as a cited source in the Overview itself. Clicks from cited sources inside Overviews tend to be high-intent; the reader chose to go deeper, which is a different kind of visitor than a quick-answer seeker.
Track restructured pages week-over-week in GSC. Look at CTR changes on the specific queries you optimized for. Give it four to six weeks before drawing conclusions.
Where to Start This Week
If you haven’t addressed AI Overviews yet, here’s the sequence I’d use:
- Pull your top ten informational pages by impressions from GSC and identify which queries now trigger AI Overviews.
- Rewrite the opening paragraph of each page to lead with a direct answer in one to two sentences.
- Add a FAQ section (five to seven questions) to each page and implement FAQPage schema.
- Validate your Recipe and Product schema with the Rich Results Test—fix anything flagged as an error before moving on.
- Audit your internal links: does every informational page link to at least one product or conversion page, and does that page link back?
That’s a week of focused work, not a quarter-long initiative. The pages most at risk from AI Overviews are already losing clicks—restructuring them pays for the effort quickly.
Frequently Asked Questions
Do AI Overviews hurt all types of business content equally?
No. Informational content is hit hardest—recipe pages, how-to guides, and definition posts see the biggest click drops. Product pages and buying guides are more insulated because AI Overviews can’t fulfill transactional intent. The user still has to click somewhere to buy, which is where your revenue is anyway.
Can I get my content cited inside an AI Overview instead of replaced by it?
Yes, and this is the real optimization target. Pages with clear structure, FAQ schema, and direct opening answers are more likely to be cited as sources. When your page appears as a cited source inside the Overview, you still get the click—often from a more engaged reader than a typical organic visitor looking for a quick answer.
Does structured data actually influence AI Overviews?
Structured data helps Google understand what your content is and how to categorize it, which supports AI Overview citation indirectly. Pages without schema are harder for the AI to parse reliably—especially for recipes and products where the structured fields are what distinguish your page from generic prose. It’s not a guarantee, but it’s a prerequisite.
Should I stop creating informational content and focus only on product pages?
No. Informational content builds the topical authority that makes your product pages trusted. The cluster approach—informational content that links to and supports product pages—is still the right architecture. What changes is how you structure that informational content, not whether you create it.
How long before restructured pages show measurable results?
Expect four to six weeks before GSC data tells you anything useful. Schema changes take time to be picked up and reprocessed. Track CTR on specific restructured queries rather than aggregate traffic, which will mask movement at the query level where the real signal lives.
If you’re running a content operation across multiple sites and want to handle this kind of restructuring systematically rather than page by page, the rest of this site covers the lean, largely automated approach I use to manage that kind of work at scale without adding headcount.