Start Date

11-12-2016 12:00 AM

Description

Implicit recommendation agents (RAs) are software artifacts that identify the interests or preferences of individual users based on preexisting consumer data and accordingly make personalized recommendations to those users. Implicit RAs are widely used by firms such as Netflix and Amazon. In this study, we analyze the seemingly paradoxical relationship between the decision effort minimization that is afforded by implicit RAs, and the effort that is required on behalf of users in order to achieve effective persuasion. We propose a persuasion model that is based on construal-level theory and the elaboration likelihood model. We tested the model in an experiment, and found that a match or a mismatch between users’ construal of a goal, and the abstractness (or concreteness) of a personalized recommendation, can influence the extent to which personalization is effective. We also found that under certain conditions, effective personalization should be based on diversity rather than on similarity.

Share

COinS
 
Dec 11th, 12:00 AM

A Construal-Level Approach to Persuasion by Personalization

Implicit recommendation agents (RAs) are software artifacts that identify the interests or preferences of individual users based on preexisting consumer data and accordingly make personalized recommendations to those users. Implicit RAs are widely used by firms such as Netflix and Amazon. In this study, we analyze the seemingly paradoxical relationship between the decision effort minimization that is afforded by implicit RAs, and the effort that is required on behalf of users in order to achieve effective persuasion. We propose a persuasion model that is based on construal-level theory and the elaboration likelihood model. We tested the model in an experiment, and found that a match or a mismatch between users’ construal of a goal, and the abstractness (or concreteness) of a personalized recommendation, can influence the extent to which personalization is effective. We also found that under certain conditions, effective personalization should be based on diversity rather than on similarity.