Application of business analytics in CRM marketing
Business analytics is the application of data science in business.
Data science is a science discipline aimed at transforming data into the new informations and knowledges. The most commonly objectives of data analysis are:
– Knowledge discovery by predicting the target variable (eg. the probability of purchase of the product X, the likelihood of churn) in the future based on the situation in the past. This approach is called supervised learning (or predictive modeling)
– Knowledge discovery in data by finding the interesting patterns (eg. to identify interesting segments and create appropriate strategies for them). This approach is called unsupervised learning.
Data science uses knowledges from several scientific fields (data science is cross-functional science). It mainly uses knowledges of artificial intelligence (data mining, machine learning), statistics, databases, IT knowledge (scripting, crawling, deployment, IT operations), and knowledges of the target (domain) area.
marketing phases of the customer lifecycle
acquire new customers
increase sales to current customers and is also the most efficient (ie. with the best benefit / cost ratio on a marketing campaign). increase in terms of purchase:
-complementary products to already have purchased product (ie. cross selling)
-more expensive variations of product that the customer bought(ie. up selling)
-rebuy of purchased product(ie. deep selling)
identification and retention customers with high probability to end relationship
recovered customes who have left
targeted marketing campaigns
Customers often prefer a simple choice of multiple choices(effect known as paralysis of decision-making), so it makes sense to offer customers a small group of products / services that are adjusted for their needs so there is high probability of interest and buy.
Alternative to targeted marketing is also called mass marketing. The main disadvantages of mass marketing are higher costs against the targeted marketing, spamming unsolicited messages which lead to furious users or customer ignorance towards the communicated messages. Targeted marketing campaigns are used in marketing in several (abovementioned)stages of the customer lifecycle. Business Analytics provides a particular methods of supporting the implementation of targeted marketing campaigns.
Application of supervised learning in Marketing
predictive modeling works by predicting the future situation on the basis of historical data.
The benefits of predictive modeling in targeted marketing campaigns for phases(mentioned above) of the customer lifecycle
-select the most appropriate prospects to reach in the various lifecycle stages in the most efficient(ie. the ratio of income / expense) way
-identification of profile characterizing those customers
-suitable communication and distribution channels
-content of communicated message for different customer groups
-prediction of customer lifetime -> use this information in determining customer lifetime value (CLV) -> use this information for targeted campaigns in multiple phases of the customer lifecycle
Application of unsupervised learning in Marketing
The goal of segmentation is to identify groups based on a certain similarities. There are several types of segmentation and it’s appropriate to combine them in certain situations together. The result of combination is highly accurate identification and description of segments.
Here are some frequently used types of segmentation:
-behavioral segmentation- identifying customer groups based on similarities in their purchasing behavior (frequency, type, sales volume ..).
-market basket analysis(segmentation of products / services) – identification of groups of products / services which customers most often purchase together.
-propensity segmentation- segmentation based on the probability of purchase of other products, probability of retirement .. (cross / up / retention models).
advantages of the shopping cart analysis in the customer acquisition and cross-selling:
-recommendation engine (eg. in e-commerce), arrangement of products on the store shelves, in leaflets
-identification right content of package suits(eg. if there are multiple items bought together by customer, then there isn’t business reason to discount both and except that promotions of that packages increase overal costs. In this case it’s reasonable to discount only one of them).
benefits of behavioral segmentation and segmentation of needs/attitudes
-design and creation of new(or adapting existing) products/services or right content of package suits adjusted for identified customer segments. eg. if customer of some segment have pleasure in some activity, then company can make some cooperation strategy with other company provided this type of activity.
-adjustment customer service based on importance and profile of this customer segments,
note: to achieve those benefits are mostly used Description of the segments and their needs / attitudes, along with marketing research
further use of segmentation b> is typically as an auxiliary method for supervised methods in a number of (the above) stages of targeted marketing campaigns. for example: segmentation identifies the profile of the segments and predictive models classify into this identified segments.
benefits of sequential modeling:
-description behavior(ie. where a user click, the sequence of pages visited, ..) of user on the web, e-shop, which can be used for:
Dynamically changed content of eShop for particular customer profile, optimize deployment of products, e-marketing (clickstream analysis for the web shop).
-analysis sequencing events in time in order to determine the most common sequence which can be used to: find the most common pathways leading to the purchase of premium products, predict the next movement within the customer lifecycle stages.