The AI Search Absorption Index: 14,728 Queries, Three Mechanisms, One Inescapable Conclusion

Overview

This study introduces the AI Search Absorption Index, a reproducible, multi-component measure of how much organic search opportunity is displaced by SERP features including AI Overviews, featured snippets, People Also Ask, knowledge panels, shopping results, and inline media.

The index decomposes absorption into three sub-indices:

• Visibility absorption — pixel-space displacement of organic results below the viewport fold
• Click absorption — estimated reduction in organic click-through rate from published CTR benchmarks
• Intent absorption — LLM-estimated degree to which the SERP satisfies user intent without a click

SERP data were collected for 14,728 queries spanning four intent categories (informational, commercial, transactional, navigational) and two devices (mobile, desktop). For queries with AI Overviews, a locally deployed Qwen 3.6-35B-A3B model assessed intent satisfaction and estimated residual organic CTR via a three-run ensemble.

Key Findings

• 67.7% of organic opportunity is absorbed: The mean composite absorption across 14,728 queries is 67.72 (95% CI [67.35, 68.09]), indicating that two-thirds of organic search opportunity is being displaced by SERP features.
• Visibility is the dominant mechanism: Visibility absorption (mean 80.02) is the highest sub-index and the primary driver in every intent category, followed by click absorption (70.03) and intent absorption (58.84).
• Informational queries are most at risk: Informational queries show the highest composite absorption (76.43), while navigational queries are the least absorbed (48.48). All pairwise category differences are statistically significant (p < 10⁻⁵).
• Mobile amplifies displacement: Mobile SERPs show consistently higher absorption than desktop (69.07 vs 66.37, p = 4.77 × 10⁻¹³), consistent with smaller viewports amplifying feature displacement.
• AI Overviews are not the whole story: The AI-only composite (41.65) is 26.07 points below the total composite (67.72), indicating that non-AIO SERP features contribute approximately 38% of measured absorption.
• Intent absorption is a distinct mechanism: Intent absorption correlates weakly with visibility (r = 0.199) and click (r = 0.241) absorption, supporting its measurement as a separate mechanism rather than a derivative of position-based metrics.

Implications

The AI Search Absorption Index provides a finer-grained view than aggregate zero-click statistics. The decomposition into three sub-indices reveals that visibility displacement is the dominant driver across all query types, while intent satisfaction operates as a distinct mechanism. For content strategists, the data confirms that informational and commercial queries face the highest displacement risk, with mobile users experiencing consistently greater absorption than desktop users.

The finding that non-AIO features contribute 38% of absorption is particularly significant: even without AI Overviews, SERP crowding from featured snippets, PAA, and inline media substantially reduces organic opportunity. This suggests that AEO strategies must address the full SERP feature landscape, not just AI-generated answers.

Methodology

Dataset: 14,728 unique keywords validated for US English search volume, distributed across four intent categories and two devices (mobile, desktop). SERPs collected in June 2026 via DataForSEO Standard SERP API with SerpAPI validation.

Feature detection: Nine SERP features detected via structured JSON parsing — AI Overview, featured snippet, People Also Ask (count), knowledge panel, shopping results, local pack, inline videos (count), inline images (count), short videos (count). No visual inference used.

Visibility absorption: Calculated as Σ(feature_pixel_height × click_impact_weight) / viewport_fold_height × 100. Feature weights calibrated by expected click-stealing impact (AI Overview: 0.90, featured snippet: 0.60, inline videos: 0.50, PAA: 0.40, etc.). Viewport fold height set to 600px (mobile-first).

Click absorption: Estimated using published CTR benchmarks and a cumulative stacking model with diminishing returns (logistic transform for raw_penalty > 60). Baseline position-1 CTR: 2.6%; with AI Overview: 1.0%.

Intent absorption: For queries with AI Overviews, a locally deployed Qwen 3.6-35B-A3B model (temperature 0.7, top-p 0.8, 262k context) made two independent assessments: intent satisfaction (Yes/Partially/No) and estimated residual CTR. Each assessment run three times; satisfaction resolved by majority vote, CTR by median. Formula: 0.60 × satisfaction_score + 0.40 × ctr_component.

Composite score: Sub-indices combined using category-specific weights. Default weights: click 0.45, intent 0.30, visibility 0.25. Informational override: click 0.35, intent 0.45, visibility 0.20.

Validation: LLM self-consistency measured via 3-run ensemble (median absolute deviation: 4.2 points for intent absorption, 6.8 points for estimated CTR). Human validation protocol proposed but not yet conducted.

Data and Code Availability

The full dataset, code, and interactive browser are published alongside this paper to enable replication and extension:

• Full dataset: absorption_index_dataset.csv and .json (14,728 rows)
• Category summary: absorption_index_category_summary.csv and .json
• Interactive browser: absorption_index_browser.html
• Pipeline code: main.py and steps/ directory
• Configuration: config.yaml