The Effect of Recommendation Framing on the Outcomes of Recommendation Agents

Abstract

E-commerce platforms offer product recommendations according to various recommendation algorithms. This research explores how businesses should frame the ways they derive their recommendations to achieve higher clickthrough rates and the perceived usefulness of the recommendation agent. For the same recommendation, companies can alter their framings based on the type of recommendation agents, from a baseline framing (e.g., “Recommended product”) to item-based framing (i.e., similarities among products) and user-based framing (i.e., the similarity between customers). Our preliminary results show that framing the same recommendation as item-based or user-based (vs. baseline framing) will increase the clickthrough rates and perceived usefulness of the recommendation agent. Meanwhile, user-based framing (vs. item-based framing) increases the recommendation agent’s perceived usefulness but does not trigger a difference in clickthrough rates. It is because the user-based framing matches both product similarity and shared tastes with other customers. Contributions are also discussed.