Google AI Overview Holiday Shopping Test: The 57% Pullback That Changes Everything

Data Collected: Analyzed AI Overview presence patterns across major ecommerce categories to understand:

• Daily percentage changes in AI Overview coverage
• Category-specific retention patterns
• Search volume correlation with AI Overview presence
• Year-over-year pattern changes
• Keyword intent and funnel stage analysis

Key Finding: Google retained only 30% of

AI Overviews at peak, with dramatic differences by category. Grocery maintained 56% retention while Furniture dropped to 3%—revealing a deliberate strategy to deploy AI where it adds value without disrupting commerce.

The September Spike Pattern

Coverage Timeline

• Sept 1-10: 9% baseline coverage
• Sept 11-18: Surged to 26% (peak on Sept 18)
• Sept 19-30: Rapid pullback to 11%
• Oct 1-15: Stabilized at 9%

The Magnitude Shift

• 2024: Gentle 5.6% reduction (321 → 303 keywords)
• 2025: Massive 56.8% reduction (741 → 320 keywords)
• 10x larger pullback year-over-year

The Category Hierarchy

Winners: High Retention Categories

Grocery & Food - 56% Retention

• Recipe queries maintained strong presence
• Ingredient information valued by AI
• Food preparation guidance prioritized

TV & Home Theater - 43% Retention

• Comparison content survived cuts
• "Best TV for [use case]" queries retained
• Technical specification explanations kept

Small Kitchen Appliances - 37% Retention

• "How to use" content preserved
• Product comparison queries maintained
• Feature explanation content retained

Losers: Low Retention Categories

Furniture - 3% Retention (97% removed)

• Visual shopping experience prioritized
• Traditional galleries preferred
• Limited informational value for AI

Home - 7% Retention (93% removed)

• Decorating queries removed
• Shopping-focused vertical
• Visual browsing emphasized

Apparel - 23% Retention (77% removed)

• Fashion requires visual discovery
• Size/fit better served by reviews
• Brand shopping preserved for traditional search

The Search Volume Revelation

2025's Complete Strategy Reversal

Unlike 2024, Google now retains higher-volume keywords:

• Retained keywords: 13,675 median search volume
• Removed keywords: 12,817 median search volume
• Ratio: 1.07x higher volume for retained

Volume Distribution

Removal rates were surprisingly uniform across volume quartiles:

• Q1 (Lowest): 71.5% removed
• Q2 (Med-Low): 73.1% removed
• Q3 (Med-High): 75.0% removed
• Q4 (Highest): 66.5% removed

The Intent Pattern

What Google Kept

Middle-Funnel Dominance

• 26.3% of retained keywords are evaluation/comparison queries
• "Best [product]" queries show 25% retention
• "X vs Y" comparisons maintained strong presence

Research & Learning

• "How to" queries retained where applicable
• Educational content about products preserved
• Comparison and evaluation prioritized

What Google Removed

Bottom-Funnel Purge

• Transactional keywords heavily removed
• Price-related queries eliminated
• Specific product names dropped
• "Buy" and "deals" queries removed

The Strategic Logic: AI Overviews help during research, step back during purchase

Holiday Predictions Based on Patterns

Expected November Behavior

If 2025 follows 2024's seasonal pattern:

• Current: 9% coverage
• November projection: 10-11% coverage
• Rationale: Research phase intensifies

Expected December Behavior

• December projection: 8-9% coverage
• Rationale: Purchase intent dominates
• Pattern: AI steps back for shopping season

The Opportunity Window

• November: Citation opportunities during research phase
• December: Traditional search dominates purchases
• Critical timing: Content must be ready NOW

Strategic Implementation Framework

For High-Retention Categories (>40%)

• Double down on comparison content
• Create comprehensive buying guides
• Build "best of" content for every segment
• Focus on educational material

For Medium-Retention Categories (20-40%)

• Test both AI and traditional optimization
• Monitor weekly for pattern changes
• Create topic clusters for stability
• Balance informational and transactional

For Low-Retention Categories (<20%)

• Prioritize traditional SEO tactics
• Focus on shopping feed optimization
• Invest in visual content
• Maintain product grid prominence

Universal Patterns Across Verticals

Despite category differences, certain patterns hold:

Content Types That Win

• Evaluation content ("best [product for use case]")
• Comparison content ("X vs Y")
• Educational content ("how to use")
• Higher search volume queries

Content Types That Lose

• Transactional queries
• Specific product searches
• Price-focused content
• Brand-specific queries

The 82% Reshuffling

Year-over-year keyword overlap: Only 18%

This massive reshuffling indicates:

• Google is actively experimenting
• Strategies must be flexible
• Historical performance doesn't guarantee future presence
• Monitoring is more critical than ever

Actionable Insights for SEO Professionals

Immediate Actions

1. Audit your content: Identify evaluation vs transactional pages
2. November priority: Get comparison content indexed NOW
3. Category check: Assess your vertical's retention rate
4. Volume analysis: Focus on 13K+ search volume keywords

Long-Term Strategy

1. Dual approach: Win AI for research, traditional for transactions
2. Content clusters: Build comprehensive topic coverage
3. Monitor volatility: Weekly tracking through holidays
4. Prepare for change: Only 18% YoY consistency

Technical Methodology

Data Sources:

• Daily AI Overview tracking Sept 1 - Oct 15
• Keyword-level analysis across ecommerce categories
• Search volume correlation analysis
• Category-specific retention calculations

Measurement Period: September 1 - October 15, 2024 and 2025

Analysis Tools: Manual tracking and proprietary monitoring systems

Key Takeaways

1. The 57% Rule: Google pulled back 57% from peak, 10x more aggressive than 2024
2. The Category Split: 56% retention (Grocery) vs 3% (Furniture) shows clear priorities
3. The Volume Flip: Higher volume now retained (opposite of 2024)
4. The Intent Filter: Research queries win, transactional queries lose

The 82% Shuffle: Only 18% keyword overlap year-over-year