The logic to send the offers is customizable. But in this case, we’re sending the offer 1) if the customer stays close to a specific product for a certain amount of time, or 2) if they look at the product, leave, and then come back to look at the product again a certain number of times. In either case, you can define the application to only send the offer if you have sufficient inventory.
The application itself has a dashboard that allows you to track the sales per store and look at the sales traffic over time. You can see how it varies depending on the time of day. It also tracks the types of items actually being sold and the offers that have been made to customers in real-time. In addition to this chart, each offer will result in an alert that is sent to the dashboard. In that alert, you can see profile of the customer that was sent the offer, which store they were in, and what offer was made.
If you drill down to a particular store, then you can see a heat map of where the customers actually are in real time. You can see which aisles are trafficked the most, and how that varies over time. You’ll also see the sales for that particular store over time and what offers were actually being made to customers in that particular store. Crucially, the dashboards also show the impact of these these real-time offers make in how much customers actually spend.
When looking how this is built, all the processing is through streaming data flows. We have sources at the top for the beacon information and the sales data pulled from a databases and change data capture. The inventory information, products, and customers are loaded into in-memory caches so they can be joined in real-time. If we look at one of the processing flows, you can see the process is done through a query. This is the query that’s checking for customers staying in one place for too long, or going back to the same place multiple times. If either those queries are met, we might make the customer an offer, but first we need to check to see if the inventory is sufficient. You can see that twenty percent mark there.
The dashboards built on this data flow are fully customizable. Each visualization is powered by a query and configured through a very simple drag-and-drop configuration enabling you to map the query to the visualization.
If you would like a more in-depth look at this application, please request a demo with one of our lead technologists.