Likes Versus Views: How Social Cues Shape Purchase Intent in Mobile Commerce

In the third and largest of three experiments, 209 participants were shown fruit recommendations on a mobile-style screen. Each saw one of four versions: likes from friends, likes from the crowd, views from friends, or views from the crowd. What they said about their intention to buy depended less on the fruit than on which pairing they were shown.

The work comes from researchers in the Department of Digitalization at Copenhagen Business School in Denmark and the School of Business at Renmin University of China. They set out to examine what they call social paratexts: the small elements that surround a piece of content, such as likes, views, emojis, or follower counts. On a phone, where screen space is scarce, these cues carry a lot of weight. Earlier research had looked at social recommendations in broad terms, but few studies had asked how likes and views differ psychologically, or how each interacts with social distance, meaning whether the feedback comes from friends or from the crowd.

The team built on Construal Level Theory, which holds that people think differently depending on psychological distance. Close events tend to be seen in concrete, detailed terms, a style called low-level construal. Distant events tend to be seen in abstract, generalized terms, or high-level construal. The researchers proposed that likes signal deeper engagement, clearer attitudes, and a stronger emotional connection, and so fit low-level construal. Views reflect more passive engagement and general popularity, and so fit high-level construal. From there they predicted that persuasion would be strongest when the cue matched the source: likes from friends should beat likes from the crowd, and views from the crowd should beat views from friends.

The first experiment, with 236 participants, used the Implicit Association Test. People were faster to link likes with specific, concrete concepts and views with broader, abstract ones. The second, with 129 participants in a simulated mobile recommendation setting, measured immersion, clarity of attitude, and how abstract the impression felt. Likes produced higher immersion, clearer attitudes, and more specific impressions. Views were tied to vaguer, more generalized perceptions.

The third experiment, a 2×2 design, tested the prediction directly. The interaction was significant. Likes from friends produced higher purchase intentions than likes from the crowd, while views from the crowd outperformed views from friends. The authors read this as confirmation that a match between paratext type and source raises persuasive impact.

The authors argue that this adds to Construal Level Theory by identifying social feedback cues as new triggers of construal level, alongside factors such as temporal framing, mood, and sensory cues. For platforms, they offer practical guidance. Likes suit detail-rich, personal, or emotionally engaging products, especially when friends endorse them. Views suit trend-driven, broad-appeal products. Their examples: a niche recipe app might show likes from friends who tried specific dishes, while a mass-market gadget might show high view counts across its user base. They also suggest pairing likes with near-term, actionable offers and views with long-term, aspirational campaigns, and tailoring displays to users, for instance showing likes from same-gender friends to leverage closeness.

The authors acknowledge limits. All three studies took place in controlled lab or online settings, which may miss real-world complexity. Every participant was a Chinese social media user, so cultural factors such as collectivism versus individualism could change how likes and views are read. They call for field experiments and live platform data, cross-cultural replication, and study of other cues such as comments, shares, and favorites.