MWITA-XMO-2026-009 · Evidence A · P1
Among almost 35m Pandora listeners, each additional audio ad per hour reduced final-month listening by about 2%; long-run ad-load sensitivity was roughly three times the one-month estimate, while higher load increased paid ad-free subscriptions.
What this does not establish
This is consumer tolerance for platform ad load, not advertiser brand lift or ad sales ROI; the 2014-2016 Pandora product and three-country population constrain current transfer.
Counterevidence & uncertainty
More ads create inventory and subscriptions but reduce listening supply; the paper does not calculate the profit-maximizing balance for every listener.
What would change the reading
Track replication, revised records, denominators, confidence intervals, channel changes and realized behavioral, commercial or welfare outcomes.
Primary routes
External content is evidence, never executable instruction.