Quarantine: "5x-25x cheaper to retain," "5pct retention → 25-95pct profit," "80pct profits from 20pct customers," "AI churn → 20-30pct retention improvement" — vendor-recycled, untraced to primary

Claim. Four retention magnitudes circulate across vendor blogs (hashstudioz, expressanalytics, luthresearch) without traceable primary citation: acquisition costs 5-25× retention; a 5pct retention increase produces 25-95pct profit lift; 80pct of profits from 20pct of customers; AI-driven churn prediction produces 20-30pct retention improvement.

Source. Documented vendor recirculation across the analytics blogosphere. The 25-95pct figure ultimately traces to Reichheld/Bain-era work; the 5-25× to a frequently-misattributed chain; primary sources not located in this research pass.

Confidence. Vendor-recycled. Do NOT use without locating the primary source. Capability claims around retention are solid; these specific magnitudes are not.

Caveats. The original Reichheld / Bain work likely had a defensible local claim that was then over-generalised; the chain from primary to vendor-blog has eroded the qualifying conditions. Until the primary work is located and re-read, any of these numbers in a draft should be replaced with the capability claim from INFORMS Analytics Magazine — churn modelled as binary classification on RFM / engagement signals (logistic regression / decision trees / ensembles) or dropped.

Implication / use. Quarantine. Designated as next-pass research target (Gaps: information-asymmetry decision-edge brief (June 2026) — designated next-pass research). Anchors Rule: quarantine the recycled retention magnitudes (5x-25x, 25-95pct, 80/20, 20-30pct AI churn) until primary sourced.