Contact US button
Successes - Case Studies - Content
Meet few of our customers
Developing web enabled dashboard
Business challenge is that customer level information is spread across various databases and transaction processing system. Other challenge is that there is no standardized MIS being generated as a single source of truth. Every unit/ function is maintaining its own MIS resulting in duplication of effort and difficulty in re-conciling numbers across functions.
Information is consolidated across all source systems, data variables definitions are standardized, list of KPIs is agreed upon with stakeholders, MIS generation is automated and KPIs tracking is hosted on intranet serving as single source of truth.
Cross Sell model
Response rate on cross-sell campaign of selling personal loan product to Credit card customers is coming down. The challenge is to increase the response rate and bring down the cost of marketing campaign.
Propensity model is built using customer's demographic data, transaction data and response data of previous campaigns. This model predicts the customer's likelihood of buying this product. Model gives a significant lift of 50% in response rate by focussing on only 40% of the customers. Net interest income can be increased by 15%.
Increasing DM (Direct Marketing) response rates for a retail chain
A key challenge for a big retail chain is to maximize the response rate of every direct marketing campaign and maximize the ROI of marketing investment.
Customer segmentation and profiling of the customer base is done using hybrid model that overlays transaction data on demographic data. Using response data of previous campaigns and logistical regression technique ,propensity models are built for various campaigns. Targeted offers are sent to specific segments using the propensity models. This results in increasing the response rates by 30% to 40%.
Developing early line management strategy to cut credit losses
Business challenge is to develop an early intervention strategy for card portfolio in first 30 days to cut credit losses. Standard behaviour scorecard based interventions come into effect only after 6 months of transaction data.
A decision tree is developed to identify the discriminating variables based on the first 30 days of transaction data. Key variables are identified which predict the customer's likelihood of going bad using decision tree tool . An intervention model is put in place basis these findings. This results in 9% to 12% betterment on NCL for 1st year.
Developing portfolio management strategy for credit cards
Challenge is to arrest the declining revenues because of increasing multi-carding and waning annual fees. Therefore business needs a segmentation framework and landscape of intervention strategies to build revenue momentum.
Cards portfolio is segmented based on revolve rate, balance and risk score and segment specific intervention strategies are developed to drive balances and fee income. Various response scorecards and pricing models are developed to cross-sell multiple products to card customers. This results in achieving 15% to 20% increase in revenue per customer over 24 months.
Save offers given to a card customer at the time of attrition are rule based and do not take into account the customer profitability. The challenge is to design a retention model which increases the save rates and optimize the NPV of saves.
A response model is built which predicts the likelihood of a customer being saved post the offer being given. An offer matrix is designed which gives the discount or an offer that the customer would be given basis the last 12 months profitability of the customer. This is overlaid on the response scorecard to decide customer level offers. This model results in 10% increase in save rate and 10% to 15% increase in NPV of saves.