DATA MINING AND STATISTICAL ANALYSIS

Churn modeling

It is well known that the price to acquire new customers is 5-10 times more than the cost of retention. Our churn models help companies to predict which customer will churn, i.e. terminate service contract, stop buying company’s products and etc., when he will churn, and what factors are affecting the decision of the customer to leave. The outcome of the modeling is as follows:

• Predictive model that assigns probability to leave (scores) to customers. Thus, the company can have the list of customers with high probability to churn. The company then has a chance to target those customers with retention strategies.
• Factors affecting churn – Model makes apparent the most important factors affecting churn. Retention strategy of the company will target mitigation or elimination of those factors.

Churn modeling is done using different statistical techniques such as Logistic Regression, Discriminant analysis, Classification trees, Survival analysis, etc.

Time Series forecasting

We will build models that can predict demand, sales, expenditures and other important metrics over time.
Our models help clients to have better understanding on what is going to happen in the future, what are the determinants of the change
(seasonality, trend, random fluctuations, etc.). 

Affinity Analysis

Affinity Analysis (Market Basket analysis, Association rule mining, etc.) is a technique used to discover occurrence of activities performed by a specific individual. For example, analyzing large database of sales transaction data allows the retailer to understand which products are bought together. Basically, Affinity Analysis helps the company quantitatively understand buyer behavior. Affinity analytics is used in the following spheres:

• Effective store-layout design,
• Cross-promotional programs,
• Recommendation systems,
• Cross-selling and up selling.

In essence, the output of the Affinity Analysis is a set of rules with some measure of interestingness, such as If the customer had bought milk and bananas, he will buy ice-cream with the probability of 56%. The set of rules are later either used by the company in developing marketing strategy or are deployed into company’s IT infrastructure as a product recommendation system. 

Customer lifetime value analysis

One way to analyze the efficiency of acquisition strategy, estimate marketing and define target segments is to calculate the Lifetime Value (LTV) of a customer. Broadly defined, LTV is the projected revenue/profit that a customer will generate during their lifetime. Customer Lifetime value analysis helps the company to solve the following problems:

• Define the most and the least profitable customer segments
• Optimize retention and acquisition strategies targeting the most profitable customer segments,
• Find the best communication channels for the given customer segment,
• Determine the optimal level of marketing investments for every market segment.
• Keep focus on long-term customer value,
• Determine the factors affecting customer lifetime span and reduce churn.

Customer segmentation

Segmentation is an analytical technique that helps the company to segment customers in a market.

With segmentation, customers are classified into homogenous groups or segments such that each group of customers shares common characteristics in a way that makes it viable for the firm to design specific offerings, products, promotional, retention campaigns for selected segments. Segmentation is usually done based on customer demographics, behavioral data, customer needs/attitudes, loyalty, etc. Sometimes segments are developed on the mix of above mentioned data. Segmentation helps the company with the following:

• Better understand the market,
• Identify different segments in a market,
• Understand what are the determinants of customer segments,
• Choose the most attractive customer segments and design effective marketing campaigns.

The data source for the segmentation can be company’s internal CRM database and/or results of specifically designed surveys.

Customer Satisfaction studies

Satisfaction is the key to customer loyalty.  Customer satisfaction survey (CSS) helps the company to understand if the services or products delivered by the company meet customer expectations. CSS allows the company to see its strengths and weaknesses, identify steps in customer journey that makes customer unhappy and leave the company. Based on the results of CSS, Customer Satisfaction Index (CSI) is calculated. The CSI is seen as a key performance indicator for a company. Our survey help to track CSI over time, understand which departments, service points, service/product attributes contributes to the satisfaction the most, etc. Our company can provide the following deliverables.

• CSS methodology
• On-line/offline questionnaire design
• Sample size calculation and survey
• Calculation of CSS
• Revealing factors with the most effect on the satisfaction.

Conjoint analysis

 Conjoint analysis is a marketing research technique developed to help companies determine the preferences of potential customers. In particular, its goal is to determine how much customers value the different attributes of the product and the tradeoffs they are willing to make among the different attributes. Conjoint analysis is best suited for products that have very tangible attributes that can be easily described or quantified. The outcomes of the conjoint analysis are as follows:

• Prediction of the market share of a the new product, taking into account competitor’s offerings
• Determination of consumers’ willingness to pay for a proposed new product
• Quantification of the tradeoffs customers or potential customers are willing to make among the various attributes or features
• Determining the factors affecting customer choice