Case Studies

Globe Telecom: Customer Profiling Portal

By December 3, 2020 No Comments

Deriving value from customers has so far been manual and complicated. Our goal is to simplify this using AI so that businesses can eliminate friction and drive new customer initiatives, To Build a self-service interface for analysts to query customer profiles for campaign and marketing purposes

Customer Profiling Portal – A user interface for users to discover customer profiles from UUP. The search can be done based on a single attribute, multi-attribute, intent or concepts

The solution is in production for over 1.5 years and Globe now captures over 500 customer attributes which help them personalize offers, personalize interactions, help the content partners provide relevant content to the subscribers, and also understand individual customer issue real-time. Analysts will be able to segment customer profiles based on demographics and filters and use the information for campaign planning and marketing purposes.

 

The solution provides Spark-based runtime to process the OLAP style requests coming from the user. The EMR reads the data from S3 and loads into EMR Spark runtime
EMR provided a long-running Spark cluster with Job server endpoints which made a preferred option to serve a backend to a self serve discovery interface

Purpose of EMR:

The solution needs a Spark-based runtime to process the OLAP style requests coming from the user. The EMR reads the data from S3 and loads into EMR Spark runtime
EMR provided a long-running Spark cluster with Job server endpoints which made a preferred option to serve a backend to a self serve discovery interface

EMR Workflow:

The solution uses a long-running cluster using Reserved instances. The instances are of type C5. The data is stored on S3 as a daily partition. A MySQL RDS instance is used as a metadata layer. The data is ingested into S3 using S3DistCp and then processed using ETL pipelines running on AWS Glue. The processed data is then read into EMR Spark cluster.

Leave a Reply

Close Bitnami banner
Bitnami