The Key To Drive Business Performance
© Abiba Systems | [Link]
Analytics Solutions Overview
Churn Analytics Solution
Product & Technical Architecture
Deployment Model
Business Benefits
Roadmap
Pilot Project Status
Other Analytics Solutions
© Abiba Systems | [Link]
Segmentation and profiling Churn Prediction
•Identify your profitable segments and • Predict the potential churn –
their behavior individual subscriber and segment
level.
•Estimate the life time value of your
customer segments for better • Perform survival analysis for existing
marketing customers
Response models Risk models
• Identify the most sensitive customers •Identify the most risky customer
for cross selling/up selling segments and list at-risk customers
• Identify customers likely to respond • Identify the customers who are
to promotional offers going to default or delay the payment
© Abiba Systems | [Link]
• ABIBA chartered a product based approach towards analytics due to followed benefits that
accrue to the customer
• Low Total Cost of Ownership
• Low Learning curve
• Single vendor concept
• Easy deployment
• Embedded Best Practices
• ABIBA’s Analytics Suite comes as an integrated analytics product since it consists of the
following modules
• Data Capture – Captures data from the various data repositories
• Data Integration - The data is integrated and converted to the required format.
• Data Modeling – Modeling is carried out and stored
• Reporting & Analysis Module – Reports are generated based on the models built.
© Abiba Systems | [Link]
Analytics Solution Overview
• Common solution approach for different models.
• Customise based on business requirements and data handled
• Based on open architecture and component model.
• Extracts required data from customer sources, transform and load
into data marts.
• Build models using historic data
• Score the live customers for various predictions
• Presentation tool for reporting and analysis
© Abiba Systems | [Link]
An effective churn solution should help operators do the following –
1. Predict existing customer likely to churn – stop recharging
2. Predict suspended customers who are likely to recharge and come back
3. Predict the estimate future life of existing customers
Churn in scope
• Postpaid and prepaid
• Retail and Enterprise customers
• Different products and services;
• Regions and modes of Sales Channel
Churn prediction is done at
Segment level Churn prediction Customer level Churn prediction
© Abiba Systems | [Link]
Solution should predict churn and satisfy the following major customer objectives –
1. Accuracy
• Accuracy of prediction of churn when compared with actual churns of
previous time period
• Prediction Accuracy of 60% or above
2. Transparency
• Solution to identify the major drivers for churn
• Identify the key drivers for the different customer segments covered in the
project
• Present these drivers in a business user friendly interface
3. Configurability
• Configurability of solution parameters and objectives by the user team
• To support the changing marketing requirements in achieving the retention
objectives of the operators
© Abiba Systems | [Link]
Solution Highlights
UI driven Modeling
Enhanced Visualization
Scalable
Modeling Methodology
Deployment Mode Configurability
•Proven modeling technique
•Quick Deployment •Business user driven
•Modeling Expertise of Modeling –Build, Edit,
ABIBA •Consulting driven solution Maintain by Business Users
•Transparent modeling •End to End Solutions from •Managing Models - Project
methodology data processing – modeling and models
– deployment - support
•Segment level modeling •Reporting & Analysis
using micro-models
© Abiba Systems | [Link]
Product Architecture
Data Sources
Customer Churn Response Actionable Insights
Segmentation Analytics modeling
module module module
Transaction Promotion
system
Data audit Data Pre Modeling
BI
& Analysis processing & Scoring Retention
campaign
Analytics Suite
External files Product and
Pricing
Customer Intelligence Framework
Business
Objective
Insights redefining objective and data
© Abiba Systems | [Link]
Production Environment Development Environment
Function Hardware Function Hardware
ETL Server 4* Quad Core processor ETL Server 4* Quad Core processor &
32 GB RAM, 500 GB Storage 32 GB RAM, 500 GB Storage
Analysis(DB) 4* Quad Core processor Analysis(D 4* Quad Core processor &
server B) server
32 GB RAM, 1 TB Storage 32 GB RAM, 1 TB Storage
Application 1* Quad Core processor Applicatio 1* Quad Core processor &
server n server
8 GB RAM, 500 GB Storage 8 GB RAM, 500 GB Storage
Modeling 4* Quad Core processor Modeling 4* Quad Core processor &
server server
32 GB RAM, 500 GB Storage 32 GB RAM, 500 GB Storage
Specifications for a subscriber base of 3 million
For deployment of upto 8 models
© Abiba Systems | [Link]
Production Environment Development Environment
Function Software Function Software
ETL Server OS: Red Hat Enterprise Linux 5 ETL Server OS: Red Hat Enterprise Linux 5
Tools: Data Integration tool, Java Tools: Data Integration tool, Java
1.6 1.6
Analysis(DB) OS: Red Hat Enterprise Linux 5 Analysis(DB) OS: Red Hat Enterprise Linux 5
server server
Tools: PostgreSQL 8.2,Champion Tools: PostgreSQL 8.2,Champion
Application OS: Red Hat Enterprise Linux 5 Application OS: Red Hat Enterprise Linux 5
server server
Tools: Tomcat 6.x,Apache Tools: Tomcat 6.x,Apache
2.2,Champion 2.2,Champion
Modeling OS: Red Hat Enterprise Linux 5 Modeling OS: Red Hat Enterprise Linux 5
server server
Tools:Modeler,Java 1.6 Tools:Modeler,Java 1.6
© Abiba Systems | [Link]
Key Business benefits
Our Analytics solutions help to
• To identify postpaid customers most likely to churn
Analytical Dashboard
• To identify prepaid customers who are likely to stop recharging
• Multi-dimensional
To identify suspended Analytics
customers who are likely to recharge
• To identifySolution
Trending most Valuable Customers
• To identify the likely response of customers to new marketing offers
Early Warning Solution
• To identify opportunities for up selling and cross selling of services
• To identify postpaid customers more likely to default
The above predictions and analysis can be done at different segment levels –
Product, region, Age on network, ARPU category etc..
© Abiba Systems | [Link]
© Abiba Systems | [Link]
Value Proposition
• Single Vendor – One Stop shop – No need for a 3rd party analytics tools and a
separate implementation partner.
• Business user driven implementation– No dependency on analysts, a scarce
resource in many markets
• Embedded Best practices – Technology and telecom domain
• Ease of use – Customisable GUI, workflow guiding model building
• Integrated with other ABIBA solutions – solutions like loyalty management,
convergent rating
• Pre-built Telecom models – reduced time to deploy
• High level of configurability and customisation – suits rapidly changing
market dynamics
• Off-shore driven deployment – With data provided offshore
• Lower overall TCO
© Abiba Systems | [Link]
Churn Analytics Roadmap
Now March 2010
• Churn Modeling
• Integrated with database, ETL and
datamart
• Column store database
• Integrated with BI solution
ETL • Churn segment level profiling
Dashboard
• User defined reports
Canned analytics
• GIS based Adhoc reports
Dashboards and
Reports
© Abiba Systems | [Link]
September 2010 March 2011
• Campaign management interface
• Survival analytics • ENTERPRISE INTELLIGENCE AND
DECISION MANAGEMENT
• Recommendation engine
• Rule based analytics
© Abiba Systems | [Link]
© Abiba Systems | [Link]
