Executive Summary
IceMobile, an international market leader in digital loyalty campaigns within the food retail industry, has partnered with Xomnia to personalize the promotions that it offers to its users.
Using clustering algorithms and recommender engines, Xomnia provided IceMobile with solutions that take into account relevant factors when proposing a good promotion to a given client, such as the client’s transaction history, shopping behavior (e.g frequency, types of visits) and characteristics. We have designed a framework and dashboard which allows quick prototyping and shows key performance metrics for different types of algorithms. These algorithms were developed in a scalable and distributed manner to help provide food retailers with a worldwide presence.
This system gives IceMobile the opportunity to create truly personalized promotions, leading to better customer engagement and loyalty.
It has been a pleasure working with Xomnia. They are experts on the subject and very capable of explaining this knowledge to both technical and non-technical people within our company. Together we have been developing truly personalized promotions for our clients.
Thomas Hantke, Data Scientist
Case
Founded in 2002, Icemobile is an expert in digital loyalty programs, with the goal to “empower food retailers to be loyal to consumers”. Its retailer client base spans major supermarket chains like Jumbo, Coop amba, Carrefour China and others. Using the IceMobile app, users automatically get digital stamps upon making a purchase at one of the company’s retail clients, and will get a freebie or an offer upon completing a certain challenge or loyalty card.
Initially, the offers and campaigns shared with different consumers using the app were generic. All clients were sent the same promotions, including promotions involving products that a client may not be interested in or may never consume. Therefore, the company turned to Xomnia to come up with a data-driven solution to ensure that customers get promotions that suit them, based on their spending habits, shopping behavior, and characteristics (veganism or vegetarianism, allergies..etc). With this, IceMobile aims to increase the accuracy with which it targets its clients, and to avoid overwhelming or alienating any segment of their clientele with irrelevant offers.
Solution
The first solution that Xomnia’s machine learning engineers worked on with IceMobile is implementing a clustering algorithm for the app, whereby each distinctive group of consumers forms a cluster. Examples of clusters included vegetarians, or those who shopped only once per week, or those who frequently consume a given product...etc. For each cluster, a certain set of relevant campaigns is activated.
The second solution is a recommender system, which uses machine learning to create a personalized campaign for each individual user. Data that this model uses include the purchase transaction history, shopping behavior, frequency of visits, the number of products purchased per visit, and characteristics of each customer. This system follows two types of recommendations. The first type focuses on upselling, which looks for products that are key for a consumer and the frequency with which they were purchased in the past to recommend a certain type of promotion. The second algorithm focuses on cross-selling, paying attention to products that a client doesn't buy from one of IceMobile’s retail partners, and giving the client offers to convince them to try buying those products from the IceMobile app.
Impact
Based on mock testing with users, the initial user feedback was really good. It was shown that users really liked the idea of getting a more personalized experience. This, in turn, has the potential to create more user engagement and it is more enticing to customers as they get offers that apply to them.