Dynamic Segmentation of the Airline Customers Using Unstructured Data Arrays
- ~ 300 variables for more than 60 million customers
- The first implementation of a project of this type in an airline
Project Prerequisites
In order to increase the efficiency of marketing campaigns of the marketing department we had to gather and process large data volumes on our customers from various sources and perform dynamic customer segmentation to create customized proposals.
Project Purposes
Creating customer profiles based on information obtained, in order to customize commercial proposals
Dynamic segmentation of customer profiles with launching of advertising campaigns for the ready segments
Involvement of communication channels for advertising: e-mail newsletters, banners in customer accounts, SMS and push notifications
Solution Uniqueness
- Applying the Machine Learning model for preprocessing and consolidating individual customer records into a single customer profile with 99% accuracy
- Creating a structured dataset with ~ 300 variables for more than 60 million customers
Creating a uniform database from non-coordinated sources of passenger information:
- Flight and passenger data from systems of booking and reserving air tickets
- Data of airline’s website visitors
- Offline customer data
- Data storage in several unrelated databases
Customer Value
- Forecasting possible customer activity based on historical data and well-timed proposal of new services to this customer
- Automatic update of information from all trips in customer profiles
- Personalization of the mailing system. Abandoning bulk mailings in favor of customer segmentation and development of individual commercial proposals
