CASE STUDIES
Personal Loan
Background
Our client, a financial institution, had a growing customer base in the United States. Majority of these customers were liability customers (depositors) with varying size of deposits. The number of customers who were also borrowers (asset customers) was quite small, and the client was interested in expanding this base rapidly to bring in more loan business and in the process, earn more through the interest on loans. In particular, the management wanted to explore ways of converting its liability customers to personal loan customers (while retaining them as depositors).
A campaign that the client ran in the previous year for liability customers showed a healthy conversion rate of over 9% success. This had encouraged the retail marketing department to devise campaigns with better target marketing to increase the success ratio with a minimal budget.
Solution
A machine learning model was built to predict whether a liability customer would buy a personal loan or not, and identify the most significant variables that influenced the demand. An advanced marketing campaign targeting tool was developed to help the client to intelligently promote a new loan product.
5,000
TARGET ACCOUNTS
98%
MODEL ACCURACY
17%
CONVERSION RATE
31%
COST SAVINGS
Health Insurance
Background
Understanding the attributes of customers can be crucial in making business decisions. Our client, an insurance company with customers across the United States, wanted to identify the relationships between the health conditions, smoking habits, family structure, etc. of their policy holders in relation to medical claims in order to establish a comprehensive customer profile and make informed decisions on developing new products as well as determining reasonable charges for their products.
A few assumptions were made by the client and further evidence was required to prove if their assumptions were correct, which directly impacted on their decision making process as well as performance. This had encouraged the marketing department to conduct further analysis before proceeding to decision making.
Solution
A comprehensive customer profile was established and statistical evidence was provided to prove whether the given assumptions were correct or not to help the client to make valuable business decisions on developing new products and pricing their products.
1,338
TARGET CUSTOMERS
95%
MODEL ACCURACY
21%
REVENUE GROWTH
12%
COST SAVINGS
Credit Card
Background
Our client, a financial institution in the United States, saw a steep decline in the number of users of their credit card, which was a good source of income for financial institutions because of different kinds of fees charged such as annual fees, balance transfer fees, cash advance fees, late payment fees, foreign transaction fees, and others. Some fees were charged on every user irrespective of usage, while others were charged under specified circumstances.
The customer churn rate of the client was right above 16% which could possibly lead them to a loss, therefore the client wanted to analyse the data of their customers and identify the customers who would leave their credit card services and reasons for same – so that the client could improve upon those areas.
Solution
A machine learning model was built to predict whether the customer was going to churn or not, and identify the most significant variables that caused the churn. A set of actionable insights and business recommendations were generated to help the client to reduce customer churn and thereby enhance performance.
10,127
TARGET ACCOUNTS
97%
MODEL ACCURACY
11%
CUSTOMER CHURN
17%
REVENUE GROWTH
Travel Package
Background
Our client, a travel company in the United Kingdom, wanted to retain their customers for a longer time period by launching a long-term travel package. The client had launched a holiday package the previous year and 18% of the customers purchased that package however, the marketing cost was very high because customers were contacted at random without a proper marketing plan in place.
The client panned to launch a new product i.e. a long-term travel package, but they wanted to leverage previously available data to reduce the marketing cost by identifying the target segments who would purchase the long-term travel package; and any important factors that could influence their purhcasing patterns.
Solution
A machine learning model was built to predict and identify the target segments that were going to purchase the newly introduced travel package. A set of actionable insights and business recommendations were generated to help the client to improve sales and reduce marketing cost.
4,888
TARGET CUSTOMERS
96%
MODEL ACCURACY
24%
CONVERSION RATE
26%
COST SAVINGS
Used Car
Background
Our client, a budding tech startup, discovered a huge demand for used cars in the Indian market. In 2018-19, while new car sales were recorded at 3.6 million units, around 4 million second-hand cars were bought and sold. There was a slowdown in new car sales that the demand shifted towards the pre-owned market. In fact, some car sellers replaced their old cars with pre-owned cars instead of buying new ones. Unlike new cars, where price and supply were fairly deterministic and managed by original equipment manufacturer, used cars were very different with huge uncertainty in both pricing and supply. The pricing scheme of these used cars became important in order to grow in the market.
Aiming to find footholds in the used car market, the client requested to come up with a pricing model that could effectively predict the price of used cars and help the business in devising profitable strategies using differential pricing utilizing the historical and current data of 32 brands across 11 locations.
Solution
A regression model was developed to help predicting the prices of used cars and devising profitable strategies using differential pricing. A set of business insights and recommendations were provided to assist the client to understand the market further and enhance competitiveness while entering the market.
7,253
TARGET VEHICLES
92%
MODEL ACCURACY
31%
REVENUE GROWTH
0.2%
MARKET SHARE
Airline
Background
Our client, a private jet operator in Australia, aimed to determine the relative importance of each parameter with regards to their contribution to passenger satisfaction. Provided 55% of individuals who travelled using their service were satisfied. These passengers were asked to provide their feedback at the end of their flights on various parameters along with their overall experience. In the survey, the passengers were explicitly asked whether they were satisfied with their overall flight experience.
The client wanted to extract actionable insights from the flight data and customer service survey data to understand the features of their customers, predict whether a customer was satisfied with their service or not, and determine the significance of all parameters.
Solution
A machine learning model was developed to predict whether a customer was satisfied or not and identify the key parameters that played an important role in swaying a passenger feedback towards satisfaction. A series of customer and business strategies were recommended to the client for operational enhancement.
90,917
TARGET CUSTOMERS
97%
MODEL ACCURACY
71%
SATISFACTION RATIO
23%
REVENUE GROWTH
Mortgage Loan
Background
Our client, a non-banking financial company that facilitated bank-related financial services, such as investment, risk pooling, contractual savings, and market brokering in India struggled to mark profits due to an increase in loan default with a default rate of 24%.
The client aimed to determine the relative importance of each parameter with regards to their contribution as to whether a loan was going to default or not; and would like to impletement automation into their existing system for predicting whether a loan application would default or not.
Solution
A machine learning model was developed using the historical data of over 90,000 customers to predict whether a customer would default on the loan payment or not.