Vend & Go : A Mobile Payment Application with Integrated Loyalty for Vending Machines

Sammanfattning

This thesis investigates the design and evaluation of a mobile payment application with an integrated loyalty program for vending machines. The project was carried out in collaboration with Impact Solution Scandinavia AB and addresses the challenge of encouraging users to choose mobile application payments instead of established card terminal payments in fast, low effort purchasing contexts. A mixed methods approach was used. A pre-launch survey examined user attitudes toward vending machine payments, loyalty rewards and adoption barriers. A functional prototype was then developed and evaluated through moderated think-aloud user testing and post test semi-structured interviews. In addition, historical vending transaction data was analyzed to examine whether location and temporal context could predict purchased product category. The results indicated that users are primarily motivated by clear financial value, such as point based rewards, direct discounts and free products after repeated purchases. Complex registration and low perceived value were identified as the main barriers to adoption. The user testing suggested that the core navigation and checkout flow were generally understandable, but that loyalty features such as points, claims and tier progression required clearer explanation. The transaction analysis showed that location context was the strongest predictor of purchased product category, while hour of day added some information and day of week contributed little. However, the overall predictive performance was limited, indicating that contextual variables alone are not sufficient for reliable prediction of individual purchases. The study concludes that a vending payment application should minimize initial friction, make rewards immediately visible and integrate loyalty features directly into the purchase flow. Historical transaction data can support understanding of broad purchasing patterns across vending environments, but more variables are needed for stronger product category prediction.