Should managers override algorithmic pricing recommendations?
In “Human-in-the-Loop Dynamic Pricing,” CRA’s Maxime Cohen, with co-authors Liting Chen, Sentao Miao, and Yining Wang, study pricing structures in which AI systems recommend prices and managers retain the authority to accept or override those recommendations.
The paper provides a new framework for understanding when human oversight enhances pricing performance—and when overriding the algorithm can backfire.
Three key findings from their analysis are:
- Human review affects both current pricing and future algorithm learning because manager overrides change the demand signals the algorithm uses to update its recommendations.
- A hybrid human-algorithm approach can outperform both algorithm-only and human-only pricing, but only when human judgment is distinct enough to improve learning without being so biased that it undermines algorithmic value.
- The timing of human intervention matters most. Early interventions create value by improving what the algorithm learns, while later interventions are most valuable when they correct algorithmic errors.
As a result, the best outcomes are not from fully automated or fully human decisions, but from a carefully designed collaboration between the two.
