Bryan Miller

One of my favorite engineering solutions was developed using IoT, streaming analytics, and cloud-based services to derive shopper insights for a retail sporting goods business in North America. IoT WiFi beacons were installed in the stores to track the location of customers in conjunction with the customer loyalty app installed and running on their mobile devices. Using a publish/subscribe approach, the customer presence data was published and ingested into a data lake in the cloud every minute. At brief intervals the ingested date was aggregated using a tumbling window to create a temporal cohort used to determine shopper movement within the store. Additionally, customer sales history was referenced to determine which adverts should be sent to the shopper's mobile devices while in the store. The following section enumerates the components used and how they were developed: 1) WiFi beacons. The beacons were distributed throughout the store so that the shopper could be identified at any given time with a corresponding product zone. The product zones were identified with product areas such as mountaineering, team sports, camping, etc. The beacons were programmed using JavaScript and a JavaScript MQTT library to publish shopper data at one minute intervals to a cloud hosted NoSQL database. 2) Weather history and near-term forecasts were also ingested into a cloud-based data lake. 3) A cloud-based streaming analytics application was used, in conjunction with the NoSQL database, to aggregate shopper presence data in five minute tumbling windows to provide movement data and trends within the retail store. 4) A set of cloud-hosted applications were developed to send custom "offers" to the shopper's mobile device based upon the recommendations from the analytics application.