AI Overviews Hurt Blog CTR. They Barely Touch Product Search. Act Accordingly.

Two Different Google Surfaces, One Panicked Strategy

If you sell on Shopify in 2026, you have probably been pitched the same story twice. First: AI search is eating your traffic. Second: the fix is a pile of AI Overview-optimised blog posts.

Those pitches mash two Google surfaces into one panic. AI Overviews are the generative summaries that appear on many informational queries. Google Shopping and free listings are product units driven by Merchant Center data and organic shopping relevance. They do not fail - or recover - the same way.

This post separates the datasets. It follows the same honesty standard as our AI referral traffic numbers: cite the panels, label what is secondary reporting, and refuse a strategy that only works if you ignore where ecommerce money still moves.

What the CTR Studies Actually Measured

The AI Overview CTR literature is large. Three findings are stable enough to plan around.

When an AI Overview is present, organic CTR falls - often sharply - on the query sets researchers studied. Ahrefs' December 2025 update found roughly a 58% lower average CTR for the top-ranking page when an AI Overview was present, on a large informational keyword sample. Other panels land in a similar band depending on methodology and time window.

The decline was real through late 2025; early 2026 showed a partial rebound, not a full reset. Seer Interactive's longitudinal panel (53 brands, 5.47 million tracked queries, 2.43 billion organic impressions) reported organic CTR on AI Overview-present queries climbing from a late-2025 floor around 1.3% to about 2.4% by February 2026 - an 85% rise from that floor, still below AI Overview-absent queries in the same research (around 3.8%).

These studies are mostly about informational and question-shaped queries. That is the fine print the ecommerce pitch decks skip. A CTR collapse on "how does wool felt" is not the same measurement as a product query that already surfaces Shopping units.

The Ecommerce Exception

Industry tracking summarised in 2026 compilations - including Omnibound and The Stacc - reports AI Overview coverage on product / ecommerce queries around ~4%, after earlier higher readings were scaled back, while education and B2B technology queries sit in much higher bands. Treat the exact percentage as secondary synthesis of industry panels, not a Google-published official rate - but the direction is consistent across those write-ups: Google has been far more willing to summarise homework questions than to replace product SERPs.

That matches commercial common sense. Shopping ads, free listings, and product packs are how Google monetises and organises commercial intent. AI Overviews did not quietly become your PDP.

So if your blog's "best running shoes for beginners" guide lost clicks, that may be an AI Overview story. If your product listings never ranked in Shopping because the feed was dirty, that is still a Merchant Center eligibility story - and, per the Peec research we cite elsewhere, it is also the path most relevant to ChatGPT's shopping carousel.

What Still Hurts Merchants - and What Does Not

Likely hurt by AI Overviews: category blogs, comparison roundups, how-to content, non-branded informational landing pages. Expect impression/click patterns that look like "seen more, clicked less" in Google Search Console on those URLs.

Less affected on current evidence: product detail pages and Shopping / free-listing surfaces, where AI Overview trigger rates are reported far lower.

Cited-inside-an-Overview lift is real for some informational brands in Seer's work - cited brands earning materially more clicks than uncited brands on the same AIO-present queries. That is not a Geopad feature, and we will not sell catalogue scoring as "AIO citation." Citation is an off-catalogue game (authority, reviews, third-party mentions) plus extractable on-page facts. Catalogue quality helps eligibility for extraction; it does not purchase a citation.

Branded queries are a different animal again - some studies show branded CTR holding or improving when an Overview appears. Do not average branded and non-branded into one panic number.

The Budget Ordering the Data Supports

If you only remember one sequence from this series, make it this:

1. Feed and free-listings eligibility. Required attributes, price/availability match, Diagnostics triage - the free listings audit.

2. Shopping-fit titles and catalogue completeness. First-70/150 title hygiene, brand/GTIN, alt text, duplicates - the Shopping title guide and our 13 checks.

3. Collection and category pages with extractable facts. When informational CTR is compressed, durable category hubs matter more than another thin "ultimate guide."

4. Editorial content where you can win citation or branded demand. Blog posts still have jobs - proof, education, brand search - but "publish thirty AIO posts this month" is a poor default for a catalogue business.

That ordering pays in today's Shopping and organic product results whether AI Overviews expand, stall, or keep rebounding. A content pack that only pays if Google summarises your blog is the speculative bet.

How to Check Your Own Mix in Search Console

Do not outsource this to a vendor average.

Segment landing pages. Separate product URLs, collection URLs, and blog/content URLs. Watch CTR and clicks on each bucket over the same window.

Separate branded from non-branded queries where you can. A branded recovery can hide a non-branded content wound.

Look for the AIO-shaped pattern on content URLs: impressions steady or up, clicks down, average position messy. On product URLs, ask whether Shopping / free listings coverage and feed Diagnostics explain performance before you commission Overview-era blog rewrites.

Keep Merchant Center Diagnostics on the weekly list even if Search Console looks fine. GSC does not show feed disapprovals.

Frequently Asked Questions

Should I stop blogging? No. Change the job of the blog. Write for customers and brand demand; stop treating every post as an AI Overview lottery ticket.

Does schema help AI Overview citation? Structured data helps machines parse your pages and supports rich results. Causation to Overview citation is not settled. We treat schema as catalogue and rich-result hygiene - see our JSON-LD guide - not as an AIO hack.

Is Shopping "safe forever" from AI Overviews? No. Current measurement says product queries are far less Overview-saturated than informational ones. That is a planning input for 2026, not a prophecy.

Where does Geopad fit? We score and help fix catalogue fields - title quality, descriptions, alt text, and Feed Readiness - so the products those clicks used to buy are eligible and clear in the feed and on the page. We do not sell Overview citation tracking.

What should I do this week? Run the free-listings Diagnostics hour, rewrite titles on your top 30 SKUs for the first 70 characters, then re-check content URLs in GSC before buying an AIO content package.