When Every Rating Is a Five: How Well Do Consumer Reviews Capture Service Quality in Credence Markets?
(A2026-133677)

AUTHORS

Sachin Sridhar, Indian Institute of Management Bangalore; Rex Du, University of Texas at Austin

ABSTRACT

User ratings in credence services often inflate toward perfection, obscuring true provider quality. Using 73,236 verified real estate reviews linked to transaction-level outcomes, we test whether review text conveys diagnostic information beyond near-universal five-star ratings. Affective praise is abundant and predicts only faster sales. Experiential praise is common yet unrelated to price or speed and, in the case of empathy, signals worse outcomes. In contrast, rare references to credence outcomes and comparative performance reliably predict higher prices and shorter time on market. A MaxDiff study reveals a mismatch between objective and perceived diagnosticity: consumers overweight affective cues and undervalue comparative signals. Across analyses, specific textual cues, as opposed to numerical ratings, capture meaningful quality differences in credence markets.