---
type: "synthesis"
clusters: ["C1", "C9"]
id: "meta-persuasion-penalty"
sources: ["cross-day"]
---
The corpus's best cross-article empirical convergence: AI doesn't just ignore human persuasion — advanced models *penalize* overt selling. Sources converge (GEO's [[cross-day-persuasion-penalty-convergence]]): Sabbah/Acar's 16,000-choice simulation ([[concept-algorithmic-skepticism]], [[claim-traditional-marketing-fails]]), Puntoni's sponsored-tag penalty ([[claim-sponsored-penalty]]), Gale's inclusion-over-sentiment ([[claim-inclusion-is-bottleneck]]), Dubois's luxury-cue penalty ([[claim-ai-ignores-implicit-cues]]). Only price and star ratings reliably move agents ([[claim-ratings-and-price-are-universal]]); the prescription is sometimes to dial persuasion *back* ([[quote-dial-it-back]]). Structure and third-party evidence win the machine's inclusion decision; story and community win the human's final choice — you must serve both ([[cross-day-dual-audience-imperative]]). The luxury inversion tests the rule ([[contrarian-geo-backfires-for-luxury]]). Companion: [[meta-agent-as-new-customer]], [[meta-codification-imperative]].