Incentive Recommendation for Crowdsourcing Consumer Survey Using Design of Experiment

• Main Author: FADIA ADLINA (NIM : 14414010), CUT
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/26326
• Institution: Institut Teknologi Bandung

<p align="justify"> Consumer panel is a group of individual or house hold, that are being recruited by research agency to report their purchase continually. The data is used by manufacture, for example to monitor loyalty, understanding customer characteristics, and studying promo effect. In conventional operation, panel asked to record every purchase in a purchase diary and it will be collected by the agency periodically. There are some problem in the conventional operation such as high operational cost and reluctance to record purchase that results in low participation. New research agency emerge with crowdsourcing model, hereafter will be referred as crowdsourcing consumer survey. Anyone who has smartphone and internet access can participate as panel. The operation of the crowdsourcing form require panel to take snapshot of their receipt and upload it through an application. Key success of this business is crowd participation, but individual will not participate unless they are fairly motivated. <br /> <br /> Design of experiment method is used to design an effective incentive system that motivate crowd to participate. This research consist of two experiments. Experiment 1 intend to examine potential factors that influence intention to participate. 4 design factor and 8 internal factor being tested in Experiment 1. Those design factors are payment scheme, balance unit, minimum redemption, and reward variation. Experiment 1 is a 2k factorial design, so 24 combinations are tested as between-subject design. Experiment 2 intend to study reward combination that are preferred by the crowd. Reward combination consist of two factors, reward form and reward amount. 4 level of reward form are tested. Those are money, phone credit, coupon for daily needs product, coupon for lifestyle and entertainment product. 3 level of reward amount are tested. Those are Rp 5.000, Rp 25.000, and Rp 50.000. Experiment 1 is analyzed using ordinal logistic regression while Experiment 2 is analyzed using repeated measure ANOVA. <br /> <br /> Experiment 1 prove that 2 design factors and 4 internal factors has significant influence towards intention to participate. The significant design factors are payment scheme and balance unit. Payment scheme is more effective if it can accomodate all group with different average cart size. In this experiment, uniform scheme is more effective. Point is more effective as balance unit than rupiah due to its ability to disguise the actual value. Internal factors that are significant are purchase behaviour, personal innovativeness, estimated balance, and age group. Experiment 2 prove that reward amount and the interaction between reward form and reward amount are significant. In general, rewards with bigger amount are more preferred than rewards with fewer amount, regardless the form of the reward, except money. In big amount money is really preferred, but in small money is very disliked. Money - Rp 20.000 and has no significant difference with Money - Rp 25.000, both are classifed as very attractive. But when the amount is small, money is very unattrative. <br /> <p align="justify">