Google AI Overviews: How 48% of Searches Changed and What It Means for Ecommerce

Google AI Overviews have reshaped the search landscape faster than any feature Google has ever launched. From appearing on 6.5% of queries in January 2025, AI Overviews peaked at nearly 25% by July 2025, pulled back to 15.7% in November as Google refined which queries triggered them, and then surged to approximately 48% of tracked queries by February 2026. For ecommerce, AI Overviews now appear on approximately 14% of shopping queries -- a 5.6x increase from 2.1% in November 2024.

These are not abstract numbers. When an AI Overview appears, it fundamentally alters the click economics of the entire search results page. Studies show AI Overviews reduce traditional organic click-through rates by up to 61%. But brands cited within the AI Overview earn 35% more organic clicks than those not cited. Being inside the AI Overview is now one of the highest-value positions in all of search.

This guide covers how AI Overviews select sources, the specific impact on ecommerce, why traditional SEO rankings are no longer sufficient, and the optimization strategies that get your products cited.

AI Overview Presence: The Numbers

The expansion of AI Overviews has been tracked by multiple research firms. Here is the data landscape:

| Period | AI Overview Presence | Source | |---|---|---| | January 2025 | 6.5% of queries | Semrush | | July 2025 | ~25% of queries (peak) | Multiple trackers | | November 2025 | 15.7% of queries (pullback) | Semrush | | February 2026 | ~48% of tracked queries | Industry tracking | | Shopping queries (Nov 2024) | 2.1% | ALM Corp | | Shopping queries (early 2026) | 14% | ALM Corp (20.9M keywords) | | Informational shopping queries | 83% | November 2025 data | | Pure transactional queries | 13-14% | November 2025 data |

The trajectory shows Google is expanding AI Overviews aggressively across query types, with periodic pullbacks to refine accuracy. The 58% surge across 9 industries documented by ALM Corp between late 2025 and early 2026 suggests the feature is entering a new expansion phase.

Why the Pullback Happened

Google reduced AI Overview frequency between July and November 2025 after quality concerns. AI Overviews were appearing for queries where they added little value (simple navigational queries) and in some cases provided inaccurate information. The pullback was Google refining the trigger criteria, not retreating from the feature. The subsequent surge to 48% confirms this -- Google recalibrated and then expanded more aggressively than before.

How AI Overviews Select Sources

The source selection pipeline for AI Overviews is fundamentally different from traditional organic ranking. Understanding this pipeline is the key to visibility.

The Multi-Stage Filtering Process

Google's AI Overviews select sources through a progressive filtering pipeline that narrows 200-500 candidate documents down to 5-15 cited sources:

Stage 1: Semantic Retrieval. Google's search infrastructure retrieves candidate pages that are semantically relevant to the query. This is the widest net -- hundreds of pages from across the web index.

Stage 2: Query Fan-Out. The original query is split into multiple related sub-queries. If a user searches for "best wireless earbuds for running," the fan-out might generate sub-queries for "sweat-proof earbuds," "earbuds that stay in during exercise," "wireless earbuds battery life comparison," and "earbuds ambient sound mode." Pages that appear across multiple fan-out queries score higher.

Stage 3: E-E-A-T Authority Filtering. This functions as a binary pass/fail gate. Pages must demonstrate Experience, Expertise, Authoritativeness, and Trustworthiness to proceed to the next stage. Pages without clear author credentials, on domains without established topical authority, or with thin content are filtered out entirely.

Stage 4: Gemini LLM Re-Ranking. The remaining candidates are re-ranked at the passage level by Google's Gemini model. This is where passage-level extractability becomes critical -- Gemini evaluates whether specific passages can be cleanly extracted and cited in the overview.

Stage 5: Data Fusion. Selected passages from multiple sources are fused into a coherent summary with inline citations. The AI Overview is constructed from the best passages across multiple pages, not from a single winning page.

The Collapse of Organic Ranking as a Proxy

The most important finding for ecommerce merchants: only 38% of AI Overview-cited pages now rank in the organic top 10, down from 76% less than a year ago. This means traditional SEO rankings are becoming an unreliable predictor of AI Overview visibility.

The breakdown of where citations come from:

38% from organic top 10 -- These are the pages where traditional SEO and AI Overview visibility overlap
31.2% from positions 11-100 -- Pages that rank on pages 2-10 of Google but still get cited in AI Overviews
31% from beyond the top 100 -- Pages that do not rank in Google's top 100 for the query at all, including 18.2% from YouTube URLs

This data destroys the assumption that "if you rank well in Google, you'll appear in AI Overviews." A page at position 47 in organic results can be cited in an AI Overview while the #1 organic result is absent. The selection criteria are fundamentally different.

What Makes a Page Citable

Pages that earn AI Overview citations share specific structural characteristics:

1. Passage-Level Extractability. The AI Overview pipeline extracts self-contained answer units of 134-167 words. If your content cannot be meaningfully extracted at the passage level -- because answers span multiple paragraphs, require context from other sections, or are buried within marketing copy -- the system moves to the next candidate.

Optimal structure: each H2/H3 section should begin with a direct, self-contained answer to the section's topic in 2-3 sentences, followed by supporting detail. The opening sentences should make sense without any of the surrounding context.

2. Entity Density. Pages with 15+ Knowledge Graph entities per 1,000 words are cited significantly more often. For ecommerce, entities include: product names, brand names, model numbers, specific prices, technical specifications (wattage, weight, dimensions), certification names, competitor product names, and standard names.

3. Multimodal Content. Pages that combine text with tables, images, and structured data perform better than text-only pages. The AI Overview pipeline can extract and reference data from tables, making structured comparison data a strong citation driver.

4. First-Hand Experience Signals. Content that demonstrates actual experience with a product -- original photography, specific test results, hands-on observations -- passes the E-E-A-T gate more reliably than content that synthesizes information from other sources.

The Ecommerce Impact: Traffic Redistribution

AI Overviews are redistributing search traffic in ways that create both winners and losers in ecommerce.

Traffic Loss Patterns

AI Overviews reduce traditional organic CTR by up to 61% for queries where they appear
Informational shopping queries see the highest AI Overview frequency (83%), meaning research-phase traffic is most affected
Sites that previously ranked in positions 3-10 for informational queries are experiencing the steepest traffic declines, as AI Overviews absorb clicks that would have gone to mid-page organic results

Traffic Gain Patterns

Brands cited in AI Overviews earn 35% more organic clicks than those not cited
The citation itself serves as an implicit endorsement, increasing both click-through rate and brand trust
Product pages cited in AI Overviews see higher conversion rates from that traffic, likely because users arrive with more context and confidence

The Zero-Click Phenomenon

For some queries, AI Overviews provide enough information that users do not click through to any source. This is the "zero-click" search outcome that has been debated since Google launched featured snippets, but AI Overviews amplify it because the summaries are more comprehensive.

For ecommerce merchants, this means:

Brand mentions in AI Overviews have value even without clicks. A user who sees your product recommended in an AI Overview may not click through immediately but may search for your brand name directly later.
Conversion-oriented content is more valuable than ever. If users do click from an AI Overview, they arrive with higher intent. Ensuring your landing pages are optimized for conversion -- not just information -- maximizes the value of reduced but higher-quality traffic.
The metric that matters shifts from traffic to revenue. Fewer clicks but higher conversion rates can produce the same or more revenue.

Optimization Strategies for AI Overviews

Strategy 1: Target the Informational Layer

Since 83% of informational shopping queries trigger AI Overviews but only 13-14% of pure transactional queries do, the highest-impact optimization targets informational content:

Buying guides -- "Best [product category] for [use case] [year]"
Comparison pages -- "[Product A] vs [Product B]: Full Comparison"
How-to guides -- "How to Choose the Right [Product Type]"
Problem-solution content -- "[Common Problem] - Which [Product Type] Fixes It?"

These pages are where AI Overviews appear most frequently and where your content can be cited. Product listing pages with minimal content rarely earn AI Overview citations.

Strategy 2: Optimize for Passage-Level Extraction

Structure every content section so the opening 2-3 sentences are a self-contained, extractable answer:

Weak (answer buried):

When we consider the various factors that impact noise cancellation performance, including driver size, seal quality, and ANC algorithm sophistication, there are several headphones worth discussing. The technology has evolved significantly over the past three years...

Strong (answer first, extractable):

The Sony WH-1000XM5 offers the best noise cancellation in its price range, reducing ambient noise by 40dB in standardized testing at 1kHz. It outperforms the Bose QC Ultra Headphones by 3dB and the Apple AirPods Max 2 by 5dB across low and mid-frequency bands. The XM5 achieves this through a combination of two processors controlling 8 microphones and a new Auto NC Optimizer that adjusts noise cancellation 700 times per second.

The second example is 67 words, within the 134-167 word extraction window, and can be cited without any surrounding context.

Strategy 3: Build Entity-Dense Content

Aim for 15+ Knowledge Graph entities per 1,000 words. For product content, this means using specific, named references instead of generic terms:

"Intel Core Ultra 9 285H processor" instead of "fast processor"
"$1,399 at Best Buy, $1,349 at Amazon" instead of "competitively priced"
"IPX5 water resistance certification" instead of "water resistant"
"Qualcomm aptX Lossless codec support" instead of "high-quality audio"

Every generic term you replace with a named entity increases your entity density and citation probability.

Strategy 4: Implement Comprehensive Structured Data

Websites with properly implemented structured data are cited in AI responses 3.2x more often than those without. For ecommerce:

Product schema on every product page with complete attributes
FAQ schema on pages with question-and-answer content
Review schema with aggregate ratings and individual reviews
Breadcrumb schema for navigation context
Organization schema for brand identity signals

Structured data does not guarantee citation, but its absence significantly reduces the probability.

Strategy 5: Create First-Hand Experience Content

Google's E-E-A-T framework prioritizes first-hand experience. For ecommerce, this means:

Original product testing with specific measurements and results
Real customer feedback integrated into product pages (not just a Yotpo widget, but curated testimonials in crawlable HTML)
Original photography showing products in actual use
Author expertise -- Product reviews and guides should have named authors with verifiable credentials in the category
Hands-on video content -- YouTube videos embedded with transcripts provide both multimodal content and an additional citation pathway (18.2% of non-ranking AI Overview citations come from YouTube)

Strategy 6: Optimize for Query Fan-Out

Because AI Overviews use query fan-out to split the original query into sub-queries, your content should address multiple facets of a topic:

A product page or buying guide for "best laptop for video editing" should cover:

Performance benchmarks (sub-query: "laptop video editing benchmarks")
GPU comparisons (sub-query: "best GPU for video editing laptop")
Budget options (sub-query: "video editing laptop under $1500")
Software compatibility (sub-query: "Premiere Pro laptop requirements")
Portability trade-offs (sub-query: "lightweight video editing laptops")

Pages that appear across multiple fan-out sub-queries are more likely to be cited in the final AI Overview.

Strategy 7: Monitor and Iterate

AI Overview optimization requires ongoing monitoring because:

Google continues to adjust which queries trigger AI Overviews
Source selection criteria evolve
Competitor content changes affect your relative visibility
Product information becomes stale

Track these metrics monthly:

Percentage of your target keywords that trigger AI Overviews
Whether your pages are cited in those AI Overviews
Click-through rates from AI Overview traffic versus organic traffic
Revenue attribution from AI Overview-cited pages
Content freshness across your product pages and guides

The New Reality: SEO + AIO Optimization

AI Overviews do not replace traditional SEO -- they add a parallel optimization surface that follows different rules. The merchants who will thrive are those who optimize for both:

Traditional SEO for the 52% of queries that still do not trigger AI Overviews (and for the organic results below AI Overviews)
AI Overview optimization for the 48% (and growing) of queries where AI Overviews appear

The critical insight is that only 38% of AI Overview citations come from the organic top 10. This means that traditional SEO, on its own, captures less than half of the AI Overview citation opportunity. The remaining 62% requires the specific optimization tactics outlined in this guide: passage-level extractability, entity density, multimodal content, structured data, first-hand experience signals, and query fan-out coverage.

Merchants who treat AI Overview optimization as a separate discipline -- with its own content strategy, measurement framework, and optimization cycle -- will capture disproportionate visibility as Google continues expanding AI Overviews across more queries and more categories.

The math is straightforward: brands cited in AI Overviews earn 35% more clicks. AI Overviews appear on 48% of queries and growing. The ROI of AI Overview visibility compounds with every percentage point of query coverage Google adds.