SearchLens 3-Bucket Classification Methodology

Input signals (per query, period-over-period)

• Position Delta (Δpos): current average position minus previous average position (lower is better in GSC).
• Impression Delta (Δimp): relative change in impressions: (imp_current − imp_previous) / imp_previous.
• CTR Delta (Δctr): relative change in click-through rate: (ctr_current − ctr_previous) / ctr_previous.
• Click Delta (Δclk): relative change in clicks.

Bucket 1 — Cited, Not Clicked (AI Overview citation theft)

Operational definition: Google likely cited or summarized the result inside an AI Overview while withholding the organic click. Rankings appear healthy; traffic does not.

IF  imp_current >= 100
AND Δimp >= -0.10      // impressions within -10% or better
AND Δctr <= -0.20      // relative CTR drop 20% or more
AND |Δpos| <= 3        // position stable within ±3
THEN → Bucket 1: Cited, Not Clicked

Agency action: Present to clients as AI Overview click interception — not a failed SEO campaign. Defend retainers with query-level proof.

Bucket 2 — Replaced by Answer (zero-click cannibalization)

Operational definition: The SERP no longer delivers measurable query-level visibility; the informational need is satisfied in-line.

IF  imp_current >= 50
AND Δimp < -0.20       // impressions down more than 20%
AND Δctr <= -0.15      // relative CTR drop 15% or more
AND |Δpos| <= 4        // position stable within ±4
THEN → Bucket 2: Replaced by Answer

Agency action: Reallocate budget (paid, brand, product-led) — not another blog refresh.

Bucket 3 — Normal Ranking Decay (classic SEO)

Operational definition: Standard ranking position loss driving click decline — unrelated to AI Overview citation patterns.

IF  |Δpos| > 3        // meaningful position loss
AND Δclk <= -0.15      // clicks down 15%+
THEN → Bucket 3: Normal Ranking Decay

Agency action: Execute on-page, technical, and authority work — the same playbook as pre-AIO SEO.

Platform limitations

• SearchLens uses read-only GSC query deltas — not live SERP scraping or Google Ads auction data.
• Bucket flags are high-accuracy pattern matches, not definitive AI Overview proof. Brief manual SERP verification is required before client delivery.
• CTR drops from paid ad density, featured snippets, or PAA expansion can mimic Bucket 1 — verify before attributing to AI.
• Query analysis pulls up to 25,000 rows per period from GSC (paginated). Extremely long-tail queries beyond that ceiling are not classified.

Health Score composite

SearchLens computes a 0–100 Client Health Score from site-wide click trend, CTR trend, average position movement, and weighted AI bucket exposure (Bucket 1 + 1.5× Bucket 2). Scores in the high-50s to mid-60s indicate recoverable AIO pressure — urgent but actionable.

Why this methodology exists

Industry studies in 2025–2026 document organic CTR degradation of roughly 34.5% to 61% on queries triggering AI Overviews. Agencies using Semrush or Ahrefs snapshots cannot isolate interception at query granularity. SearchLens exists solely to close that operational gap.