Firms regularly call on multiple social media communication tactics (e.g. influencers, user-generated content and brand messages) to interact with their customers. Drawing on both parasocial relationship and social identity theories, we investigate the relative effectiveness of brand-directed versus influencer-directed social media marketing in stimulating both customer engagement and purchase behaviour. First, we use a survey research design to collect data on customers’ intention to engage with the brand and purchase when exposed to brand-directed and influencer-directed social media marketing. Second, we collect data from a field experiment on a social media platform that captures customer engagement and brand sales at aggregate levels attributed to brand-directed and influencer-directed social media marketing. Consistently across both studies, we find that brand-directed and influencer-directed social media marketing positively impact customer engagement and customer purchase behaviour. Further evidenced across both studies, we find brand-directed social media marketing is more effective than influencer-directed social media marketing in fostering customer engagement, whereas influencer-directed social media marketing is more effective in driving customer purchases. We also find that female customers are more impacted by influencer-directed social media marketing on both dimensions than brand-directed social media marketing. The study provides a strategic direction for brands to optimally allocate their limited digital marketing budget between brand-directed and influencer-directed social media marketing based on their marketing objectives central to customer behaviour – either enhancing longer-term customer relationship building via engagement or generating shorter-term sales.
Keywords: Social media marketing, influencer marketing, customer engagement, brand sales, field experiment
- Authors: Ashish Kumar, Daniel Rayne, Jari Salo, Ching Sophia Yiu
- Journal: Australasian Marketing Journal
- Year: 2025
- Volume: 33
- Issue: 1
- Pages: 87-95
- DOI: https://doi.org/10.1177/14413582241247391
Citations: 21 (Till: 2026-05-29) (using R package rcrossref)
Recent: Google Scholar
AMJ best overall paper for 2025