USA +1 650 469 3205 | IND +91 120 4239596

Data Modeling


  • The client, a leading Financial Services and BPO firm, wanted to rationalize its database structures for long term (> 6 months vintage) warehousing perspective.
  • Based on concrete business objectives (KPI) – identify key drivers and redundancies in the database and define warehousing strategies for more than 6 months old business data.
  • Complex database structures- Backend database for multiple CRM systems (financial services, insurance distribution etc.), backend tables for the Call Centre Dialer software (message delivery reports, call logs etc).


  • For each KPI, a single snapshot* of data was created.
  • *Merged dataset for all related datasets (e.g. for Insurance distribution business call conversion rate KPI- all datasets related to the respective CRM, message / call logs from the Dialer backend were merged).
  • Conducted Key Driver Analysis (KDA) for each set of KPIs to capture features that have direct impact on the respective KPI performance/ behavior.
  • Correlations, degree of associations with the target field (related to KPI) were studied.
  • Based on these inputs and business knowledge, a cross section of features was created in order to capture as many different and relevant characteristics as possible.
  • Various feature selection algorithms (selection by regression, variable clustering, degree of association/ correlation etc. along with multicollinearity tests) were applied in order to find out the smallest set of independent and unique features that define the target value.
  • The key drivers (created fields) were reduced to actual database fields.
  • These database fields along with policy and process mandated parameters (e.g. Dialer logs in accordance with TRAI policies) were consolidated and shared with the client.


Client has a reduced DW size (>70% reduction) resulting in huge storage savings. At the same time, all the information needed for strategic and tactical purposes is still available.