Information Systems Research, 2015, Volume 26,
Issue 3, Page 552-564.
Can location-based mobile promotion (LMP) trigger contemporaneous and delayed sales purchases? As mobile technologies can reach users anywhere and anytime, LMP becomes a promising new channel. We unravel the dynamic sales impact of LMP on the basis of a randomized field experiment with 22,000 mobile users sponsored by one of the largest mobile service providers in the world. Our identification strategy is to gauge the marginal increases in consumer purchases of the geo-fenced treatment group of users who received LMP, above and beyond the baseline control groups. There are two controls: one group who received the same LMP but not in the virtual geo-fencing locational range (nongeo-fenced control), and the other who did not receive the LMP but in the geo-fencing range (geo-fenced control). The latter control serves as an organic holdout baseline from the similar population, i.e., counterfactual test of what if without the mobile LMP intervention, to identify the actual ÒliftÓ of incremental purchases caused by the treatment with the mobile LMP intervention. Findings suggest that LMP treatment has a significantly stronger impact on contemporaneous (same-day) purchases and delayed (subsequent-days) purchases than the controls. The randomized experiment design renders these findings robust to alternative explanations such as mobile usage behavior heterogeneity, product effects heterogeneity, nonrandomized sample-selection bias, and endogeneity concerns. Follow-up surveys with the field experiment users explore the nuanced mechanisms via which LMP may induce the impulsive, same-day purchases, and create product awareness for the planned subsequent-days purchases. LMP can generate six times more purchases than nongeo-fenced control with the LMP intervention, and 12 times more than geo-fenced control without the LMP intervention. LMP has a delayed sales effect for 12 days after the mobile promotions. The total sales impact of LMP could be underestimated by 54% if excluding the delayed sales impact and only including the contemporaneous impact. These findings are new to the literature and often neglected in mobile promotion practices, proffering novel implications on the sales value of LMP in the mobile era.
Keywords: mobile computing ; mobile promotion ; location-based mobile promotion ; advertising ; dynamic impact ; randomized field experiment