Going too far? How consumers respond to personalized advertising from different sources

This paper examines the extent to which the level of personalization in advertisements on social networking sites from four online sources affects source attitudes. Based on the privacy-calculus theory, the trade-off between perceived personal relevance and perceived creepiness is tested. We also examine the moderating role of source type. We used a factorial survey by setting up a 3 (level of personalization: low vs. moderate vs. high) x 4 (source type: health vs. governmental vs. commercial vs. news) between-subjects design. We tested a moderated mediation model with perceived creepiness and relevance as competing mediating variables and source type as the moderating variable based on the privacy calculus theory and social exchange theory. The results indicate that perceived creepiness negatively explains personalization perceptions. The tipping point can be found between the low and moderate level: a moderate (vs. low) level of personalization increases perceived creepiness, but high personalization does not increase it further. Contrary to our expectations, perceived relevance does not act as a positive explanatory mechanism. Finally, our findings demonstrate that source type is important: the privacy calculus for each personalization level differs for different online sources.