A large Telco provider wished to develop a price optimisation tool that would assist with their product pricing decisions with a view to optimising revenue, profit and churn related targets set for the business.Learn more
These pricing decisions need to be supported by analysis of customer product preferences, price sensitivities, competitors’ product (and pricing) offers. We built an analytical framework and supporting models to construct a tool that could automate much of the analytical processes so that pricing decisions could be optimised in real-time.
This required the application of high powered advanced analytics to continuously assess and respond to a range of factors that will ensure the client’ product offers remained competitive in a highly dynamic and competitive market place.
The Defin’d team applied its extensive machine learning IP to construct predictive models of customer choices and behaviours, the price elasticity (sensitivity) of customers and how that impacted their product choices from the client’s offer set, as well as choice between the client and its competitors.
The models use a range of disparate data sources as inputs:
- Product features (client and competitors)
- Consumer price sensitivities
- Churn rates
- Existing database
- Scenario testing
- Competitive price points
- Product usage data
Various machine learning modelling techniques (e.g. Random Forests, GBMs, etc.) were deployed to build a portfolio of linked choice and price elasticity models. The models were able to ingest regularly updated data feeds and rescore the customer database and produce projections of key financials metrics. The user could adjust portfolio pricing decisions and produce updated projections under alternative pricing scenarios.
Results: Improved marketing spend
The telco now uses the models to optimise their in-market price offers as well as gleaning insights into procurement options and commercial arrangements.