Who is gonna Churn?
Customer churn is one of the most important metrics for a growing business to evaluate. While it’s not the happiest measure, it’s a number that can give your company the hard truth about its customer retention.
It’s hard to measure success if you don’t measure the inevitable failures, too. While you strive for 100% of customers to stick with your company, that’s simply unrealistic. That’s where customer churn comes in.
What is customer churn?
Customer churn is the percentage of customers that stopped using your company’s product or service during a certain time frame. You can calculate churn rate by dividing the number of customers you lost during that time period — say a quarter — by the number of customers you had at the beginning of that time period.
Ways to Reduce Customer Churn
Focus your attention on your best customers.
Rather than simply focusing on offering incentives to customers who are considering churning, it could be even more beneficial to pool your resources into your loyal, profitable customers.
Analyze churn as it occurs.
Use your churned customers as a means of understanding why customers are leaving. Analyze how and when churn occurs in a customer’s lifetime with your company, and use that data to put into place preemptive measures.
Show your customers that you care.
Instead of waiting to connect with your customers until they reach out to you, try a more proactive approach. Communicate with them all the perks you offer and show them you care about their experience, and they’ll be sure to stick around.
Who is gonna Churn ???
The goal is provide an analysis which shows the difference between a non-churning and churning customer. This will provide us insight into which customers are eager to churn.
In this case, a bank manager is in a scenario where several customers are leaving their credit card services. It would be extremely interesting for the company to be able to predict the customers most likely to leave such services so that, in this way, the bank can act preventively in order to offer better services in favor of maintaining the customer.
The first goal of this project is to provide an analysis which shows the difference between a non-churning and churning customer. This will provide us insight into which customers are eager to churn.
The top priority of this case is to identify if a customer will churn or won’t. It’s important that we don’t predict churning as non-churning customers. That’s why the model needs to be evaluated on the “Recall”- metric (goal > 62%).
- How many customers have some attrition with the bank?
- How demographic variables are impact to earger the churn ?
- How are the relationship with the variables vs churn ?
Question1 — How many customers have some attrition with the bank ?
In order to understand how many customers have some attrition, we will look at the Attrition_Flag field.
Attrition_Flag: Internal event (customer activity) variable — Existing Customer / Attrited Customer
Results: The graph below reveals that approximately 16% of the customers present at the base have some type of friction with the financial institution. This is an important slice of analysis, given that it basically represents the target audience of customers who, in some way, are not comfortable with the services offered by this financial institution.
Question2 — How demographic variables are impact to earger the churn ?
In this first exploratory analysis session, demographic variables present in the database, such as age, gender, dependents, education, among others, will be discussed. The objective is to understand the public of this banking institution a little better and to cross these factors with other key variables that can better define possible customer migrations.
We analyze that how demografic variables which are “Age”, “Gender”, “Education Level”, “Income Category”, “Number of products bought”, “Months Inactive” impact to customer churn.
Results:
- Age variable is not impact to customer churn.
- The difference is too small to say that one gender is more eager to churn.
- The “Education level” — distribution of the churn/nonchurned customers shows no difference.
- We notice that “Income category” 60K-80K customer’s churned ratio seperate a little bit (%3) from the overall churn ratio.
- The non churned customers tend to buy more products (more than 3 products) then the churned customers.
- We notice that grader than 1 month inactivity in the last 1 year is impact to churn.
Question3- How are the relationship with the variables vs churn ?
In this question, we are trying to understand the relationship between each different variables. A correlation matrix is a table showing correlation coefficients between variables. Each cell in the table shows the correlation between two variables. A correlation matrix is used to summarize data, as an input into a more advanced analysis, and as a diagnostic for advanced analyses.
Results:
The variables Total_Trans_Ct, Total_Ct_Chng_Q4_Q1 and Total_Revolving_Bal are the top 3 features that most directly and negatively influence the churn of customers. In other words, the higher the value of these 3 variables mentioned, the lower the churn rate of these customers.
In the other analysis spectrum, the Contacts_Count_12_mon and Months_Inactive_12_mon variables are the 2 main features that have a positive correlation with the churn target variable. This means that the higher the value of these 2 mentioned variables, the higher the churn rate of the public.
Results
Our model was pretty good! It was able to yield an accuracy score of almost 89%.
The result of prediting customer churn, our model score;
- Traning Model accruracy: 90.45%
- Test Model accruracy: 89.21%
While building this customer churn predictor, we tackled some of the most widely-known preprocessing steps such as scaling, label encoding, and missing value imputation. We finished with logistic regression model to predict customer churn.
Thanks
