Complements are warm and substitutes are competent: the effect of recommendation type on focal product evaluation

Abstract

Purpose

Previous research about online recommendation systems has focused largely on their impact on customers' purchase decisions regarding the products being recommended, but it has mostly ignored how they may affect focal product evaluation. This research aimed to examine the influence of recommendation type (i.e. substitute-based vs complement-based) on focal product evaluation dependent on the brand image (i.e. warm vs competent).

Design/methodology/approach

Four laboratory experiments were conducted. Study 1 adopted an implicit association task. Studies 2 and 3 used a 2 (image: warmth vs competence) × 2 (product display: complements vs substitutes) between-subjects experimental design. Study 4 used a 2 (decision stage) × 2 (image) × 2 (product display) × continuous (need for cognition) between-subjects design.

Findings

Study 1 demonstrated a general “complementation (competition)—warmth (competence)” association. Studies 2 and 3 found that when a focal product had a warm (competent) image, complement-based (substitute-based) recommendations led customers to evaluate it more favorably than substitute-based (complement-based) recommendations. Study 3 further demonstrated that processing fluency mediates the above effect. Study 4 showed that this effect relies on heuristic processing and disappears for those who are in the screening stage or have a high need for cognition.

Originality/value

Theoretically, this research extends the understanding of the stereotype content model of focal product brand image, the feelings-as-information process, and moderating roles of processing stage and need for cognition in e-commerce contexts. Practically, the findings provide online retailers a guideline for customizing their recommendation systems.

Citation

Huang, Y., Lin, Z. and Yang, L. (2022), "Complements are warm and substitutes are competent: the effect of recommendation type on focal product evaluation", Internet Research, Vol. 32 No. 4, pp. 1168-1190. https://doi.org/10.1108/INTR-09-2020-0510