How Do Consumers Interact with Digital Expert Advice? Experimental Evidence from Health Insurance

Abstract

Consumers increasingly use digital advice when making purchasing decisions. How do such tools change consumer behavior and what types of consumers are likely to use them? We examine these questions with a randomized controlled trial of digital expert advice in the context of prescription drug insurance. The intervention we study was effective at changing consumer choices. We propose that, conceptually, expert advice can affect consumer choices through two distinct channels: by updating consumer beliefs about product features (learning) and by influencing how much consumers value product features (interpretation). Using our trial data to estimate a model of consumer demand, we find that both channels are quantitatively important. Digital expert advice tools not only provide consumers with information, but also alter how consumers value product features. For example, consumers are willing to pay 14% less for a plan with the most popular brand and 37% less for an extra star rating when they incorporate digital expert advice on plan choice relative to only having information about product features. Further, we document substantial selection into the use of digital advice on two margins. Consumers who are inherently less active shoppers and those who we predict would have responded to advice more were less likely to demand it. Our results raise concerns regarding the ability of digital advice to alter consumer preferences as well as the distributional implications of greater access to digital expert advice. This paper was accepted by Stefan Scholtes, healthcare management. Funding: This work was supported by the National Institute on Aging [Grant K01AG059843] and the Patient-Centered Outcomes Research Institute [Grant CDR-1306-03598]. The project also received financial support from Stanford Innovation Funds. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2020.02453 .

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 13 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.