Client is one of the best service providers of mobile telephony in India and has operations in India
Objective
- Standardization of 2 million records of post-paid and prepaid customer data
- House-holding
- De-duplication
- Fraud Management System
- Regulatory compliance
- Real Time Integration with the current CRM application
Benefits
- Data cleansed and standardized with 80% accuracy
- Dictionaries were built for 31 tokens of addresses and names
- Data is standardization and parsed upto 31 tokens so that house-houlding can be done based on any/combination of these tokens
- House-holding will enable marketing group to
- Perform micro-level customer data analysis
- Identify new customer segments
- Build strategies to acquire new customers
- FMS solution will enable fraud management and will ensure no blacklisted customer gets a subscription without an approval
- House-holding to analyze the number of connections each customer has in order to provide such report to regulatory authority
- Solution will enable Recovery department in
- Address verification
- Bill delivery and recovery
Solution
Spice Telecom evaluated and selected Business Objects Data Quality (DQ 11.5) product and chose IIPL as implementation partners.
IIPL conducted workshops and met business users, IT personnel to collect data on envisaged business challenges and problems related to Data Quality.
IIPL conducted a two-week requirements gathering exercise in order to understand business requirements and helped client define business rules for the data. After which, the project was implemented as per IIPL’s Data Quality Implementation Methodology. As part of which, IIPL accomplished the following activities,
- Profiled data to identify patterns, keywords, regular expressions and variations
- Developed InfoSTEP Inc. India Address Cleanse Transform (IIACT) based on the results of profiling exercise. IIACT is developed using python code and is essential as a complementary to DQ product. IIACT does parsing, cleansing and standardization of data
- Built Dictionaries for Indian Names and Addresses
- Developed the following projects on DQ products
- Standardization
- Matching and House-holding (4)
- RTI with the current CRM application
Challenges
- Profiling of 2 million records to identify patterns, variations and build standards for local addresses and names
- Too many variations in address specifications
- Different tokens can be specified similarly
- Lack of independent (external) standards for Indian addresses
- Mapping of acceptable alternatives with standard data without replicating data. E.g. Phase vs Sector, Tehsil Vs Districts
- Data did not have any delimiters and was entered haphazardly
