This study examines whether influencer credibility is associated with purchase intention through parasocial relationship (PSR) and consumer-brand engagement (CBE), and whether the proposed sequential association differs between human and virtual influencers. A cross-sectional survey of 450 social-media users in East Kalimantan, Indonesia, was analyzed using partial least squares structural equation modeling, bootstrapped serial mediation, measurement invariance testing, multi-group analysis, and predictive assessment. Credibility was positively associated with PSR (β = 0.472), PSR with CBE (β = 0.392), and CBE with purchase intention (β = 0.517). The serial indirect association was statistically supported (β = 0.096; 95% CI [0.064, 0.131]). The credibility–PSR association was stronger for human than virtual influencers (β = 0.548 vs. 0.391; permutation p = 0.008), as was the serial indirect effect (β = 0.116 vs. 0.071; p = 0.021). By contrast, the CBE–purchase intention association did not differ significantly between groups (p = 0.614). These results identify a source-type difference at the influencer–consumer bonding stage rather than at the brand-engagement–intention stage. The study contributes a dual-stage relational account that distinguishes PSR directed toward an influencer from CBE directed toward the endorsed brand. For practitioners, the findings suggest that virtual-influencer strategies may benefit from credible, consistent, and transparent persona narratives that sustain relational engagement, while both influencer types should connect audience attention to the endorsed brand. The cross-sectional, non-randomized design limits causal inference.