Roughly a third of your informational pages, and almost none of your product pages. AI Overviews appear on 36% of informational queries, 8% of commercial queries and 5% of transactional ones — a sevenfold gap between top and bottom, measured by Seer Interactive across 5.47 million tracked queries from 53 brands in its April 2026 analysis. Exposure isn't handed out by page type; it's handed out by the queries a page earns its clicks from.
That distinction cost us a rewrite. The first triage view we shipped in Contexta sorted a site's pages by folder — blog on one side, products on the other — and it was close to useless: a "best waterproof jacket" category page and a how-to post can carry near-identical risk, while two posts in the same folder differ by an order of magnitude. Rebuilding the view on the site's own Search Console queries is what made it honest, because the query is where the exposure actually lives.
Which query shapes actually pull an AI Overview?
Comparison and question-shaped queries, overwhelmingly. Seer's April 2026 data puts an AI Overview on 95.4% of comparison queries and 85.9% of question-format queries, against 36% for informational queries taken as a whole — so the risk inside "informational" is concentrated, not spread evenly.
That spread is the most useful thing on this page. Informational as a bucket sits at 36% exposure, which means roughly two-thirds of informational queries still return a plain results page as of mid-2026. The panic-grade numbers belong to specific shapes: a page living on "X vs Y" or "how do I" is effectively always sharing the screen with a summary, while a page ranking for a one-word head term faces odds closer to one in four.
Doesn't 99% of AI Overviews being informational mean your blog is finished?
No — that number describes Google's mix, not your risk. Ahrefs' April 17, 2025 study found 99.2% of keywords triggering an AI Overview are informational in intent, and it gets quoted as though informational pages are 99% doomed. It says something much narrower: of all the results pages carrying an AI Overview, nearly all happen to be informational. It tells you nothing about how many informational results pages carry one.
Those are two fractions with two different denominators. Composition (99.2%) describes what AI Overviews are made of; exposure (36%) describes your odds. Mistaking the first for the second is how an owner ends up gutting a blog that was never the problem — Google's own May 2026 optimization guidance points the same way, telling site owners they don't need a separate AI framework to begin with.
Are product and category pages really protected?
Yes, for now — and the reason should make you uneasy. Transactional queries carried an AI Overview just 5% of the time in Seer's April 2026 data, and Semrush, tracking more than 600,000 keywords across ten industries from November 2025 to April 2026, found the share of transactional results pages with an AI Overview actually fell 5% over those six months. Product pages are the one place where the trend line points down.
The protection isn't technical. Google could summarize "buy trail runners size 10" — it doesn't, because that results page is where the ads are. In the same Semrush study, Google Ads and AI Overviews now appear together about twice as often as a year earlier, and keywords carrying an AI Overview showed higher average cost per click in most industries: $5.02 against $1.51 in Jobs & Education, $4.84 against $2.14 in Finance.
Read that as a business decision, not a boundary. Your product pages are shielded by Google's revenue model, which means the shield is rented rather than owned, and the lease terms are Google's to change. It's also specific to AI Overviews, the inline summary — AI Mode is a separate surface where a shopping question gets answered conversationally and this intent split doesn't hold in the same shape.
So which of your pages should actually worry you right now?
The ones ranking for "best", "vs", "review" and "alternatives" — the commercial-investigation layer. It looks safe on today's numbers at 8% exposure, and that's precisely what makes it easy to skip: Semrush measured the share of commercial results pages with an AI Overview growing 71% between November 2025 and April 2026 — the same window in which transactional fell. Low absolute, steep slope.
The industry splits make the direction hard to argue with — Finance grew 231.25%, Computers & Electronics 107.62%, Games 76.66%. For a WooCommerce store this is the layer that feeds the shop: the buying guide, the category page ranking for "best X for Y", the comparison post people read before they ever reach a product. Meanwhile the informational hit has largely already landed. Ahrefs measured position-one click-through falling 34.5% with an AI Overview present in April 2025, then 58% in its February 4, 2026 update using the same 300,000-keyword method.
So triage forward rather than backward. The pages worth an hour this quarter aren't the ones that already bled out — they're the ones sitting at 8% exposure on a slope that has climbed for six straight months.
How do you tell which of your own pages is exposed?
Join each page to the queries it earns clicks from, then score it by the shape of those queries. Page type tells you nothing on its own: a category page ranking for "best trail runners" is a comparison query wearing a product-page costume, and it carries comparison-query risk.
- 1
Export query-by-page data, not page data
Search Console's Performance report, filtered per URL, so every page carries its real query list instead of a guess from the folder name.
- 2
Tag each query by shape
Comparison, question, commercial, transactional, branded. The words "vs", "how", "best", "buy" do most of the sorting.
- 3
Weight the page by where its clicks come from
A page drawing 80% of clicks from question-shaped queries is exposed; the same page at 80% branded is not. Most are mixed, and the mix is the score.
- 4
Hand-check the top three queries
Studies give you base rates. Typing the query yourself gives you the fact for your market and language.
- 5
Fix the climbing slope before the finished drop
Commercial-investigation pages first, informational pages that already bottomed out second.
Step one is the tedious part at scale, because a 400-page store exports into a spreadsheet nobody opens. Contexta's Problem Map imports that Search Console data and ranks pages by lost clicks per month, pulling CTR losers and page-two URLs to the top — which is where a query-shape check earns its time back. Search Console's own generative-AI report helps a little, though it reports impressions with no clicks, so it tells you where you appeared and never what it earned.
If a page is exposed, does getting cited fix it?
Partly, and this is where most advice overpromises. Being cited in the AI Overview earned roughly 120% more organic clicks per impression than not being cited, in Seer's April 2026 cohort. That same measurement puts cited pages about 38% below what the identical query returned with no AI Overview at all.
Citation is mitigation, not restoration. It's still worth earning — 120% over the alternative is real money, and the structural work behind it (self-contained passages rather than rank) makes the page better regardless. But when someone tells you that becoming the cited source wins the traffic back, the data doesn't carry them as of mid-2026.
Which is the argument for triage over a rebuild. For a page whose clicks come from transactional queries, the honest move is to leave it alone and spend the hour elsewhere. For one sitting at 95.4% comparison exposure, the honest move is to accept that the plain results page isn't coming back, and make the page good enough to be quoted inside the summary instead.
FAQ
Which queries trigger AI Overviews most often?
Comparison and question-shaped queries trigger AI Overviews far more than anything else: Seer Interactive's April 2026 analysis of 5.47 million tracked queries found an AI Overview on 95.4% of comparison queries and 85.9% of question-format queries. Informational queries as a whole sit much lower at 36%, which means the risk inside that bucket is concentrated in specific shapes rather than spread evenly. Single-word head terms trigger one about 27.3% of the time, and commercial and transactional queries trail at 8% and 5%.
Do AI Overviews appear on WooCommerce product pages?
Rarely — transactional queries showed an AI Overview only 5% of the time in Seer Interactive's April 2026 data, and Semrush found the share of transactional results pages carrying one actually fell 5% between November 2025 and April 2026. The reason is commercial rather than technical: those results pages are where Google's ads run, and the same Semrush study found ads and AI Overviews now co-appearing about twice as often as a year earlier, on keywords with higher average cost per click. Treat that protection as rented rather than permanent, because it rests on Google's revenue model and not on a technical limit.
Should I delete informational blog posts that lost traffic to AI Overviews?
No — deleting them usually removes the pages that feed your AI citations without recovering a single click. The widely quoted figure that 99.2% of AI Overview keywords are informational (Ahrefs, April 17, 2025) describes what AI Overviews are made of, not how likely your informational pages are to face one, which is closer to 36%. Rework the exposed posts into self-contained answers instead, and check the query shape behind each page before touching it.
Does ranking first still protect a page from AI Overview click loss?
No — position one takes the hit directly when an AI Overview is present. Ahrefs measured a 34.5% lower average click-through rate for the top-ranking page on keywords with an AI Overview in April 2025, and raised that to 58% in its February 4, 2026 update using the same 300,000-keyword method. Ranking still buys eligibility to be cited, but as of mid-2026 it no longer buys the click on an AI Overview results page.
