Imagine your company had implemented one of the world's most successful customer loyalty programs, with millions of customers participating.
Your program offers different tiers of membership which are reached according to the customers' activity level within a rolling, fixed time period.
Over the years the rules of your program have become quite intricate and difficult to follow. Also, there has been a shift in the way customers use your service offering and the demographics of your customer base is changing.
In short: you want to simplify the rules of your loyalty program and adjust the reward system.
The challenge is: you only have one shot in getting this right.
Any change in the program is going to affect your customers in some way or another - you must ensure that customers and their status in the membership program are not negatively affected by the changes.
Of course, you also want to make sure your bottom line isn't negatively impacted.
Thankfully, our client had a large set of GDPR-compliant historical data about their customers' service usage and about their status history in the loyalty program.
Based on that data, we set up a simulation to evaluate the potential effect of new rules, membership points and membership tiers.
The simulation helped our clients to identify a rule set and point systems that ensured all key objectives of the updated loyalty program could be met.
Be built the simulation using Mathematica and the Wolfram Language, because this environment provided us with all necessary capabilities right out of the box, making the development of the simulation very efficient and ensuring that we could later transfer it to our client easily without any extra technical dependencies.
In particular, the Wolfram computational platform provided us with:
Mechanisms to transform data that was extracted from a data warehouse and load it into the simulation.
Find patterns in the input data and simulated result data.
Setup the necessary data structures and simulation rules and equations using both statistical approaches and rule-based approaches
Run the simulation and display and export the results.
Predict future customer behavior based on probabilistic machine learning algorithms.