Churn Analysis

Churn analysis is the process of identifying and predicting customers who are likely to terminate their relationship with a company or service.

Identifying the root causes of churn: By analyzing customer data, businesses can identify patterns and trends that may be contributing to churn. For example, if a large number of customers are churning after a certain length of time, the company may need to reassess its pricing or product offerings.
Predicting future churn: Churn analysis can help businesses identify customers who are at risk of churning in the future, allowing them to take proactive measures to prevent it.
Improving customer retention: By identifying the factors that contribute to churn, businesses can take steps to improve customer retention and reduce the likelihood of churn occurring.
Increasing customer lifetime value: By retaining customers for longer periods of time, businesses can increase the overall lifetime value of each customer, which can have a positive impact on the bottom line.
Improving customer satisfaction: By identifying the root causes of churn, businesses can improve the customer experience and increase customer satisfaction, which can help reduce churn in the long term.

Added Business Value

Churn analysis is useful for a variety of purposes, including...

01.

Predicting Future Churn

This example uses a customer table to predict future churn. The table contains customer information like demographics, account details, and a column called churned that informs us if the customer has churned or not: Yes, No, or NULL (neither).

We SELECT data from the customer table. These are used as the input to the prediction model function, PREDICT .
The PREDICT function is used to predict a value for the churned column for each row in the customer table, and the ignore parameter is being used to specify that the customer_id column should be ignored in the prediction.
To look at potential future churn, we select the rows where the model predicts churn (prediction='Yes') and customer churn is undecided (churned IS NULL).
The resulting rows are then ordered by the probability column in descending order. We return all of the columns from the customer table (* indicates to return all columns), as well as the prediction and probability columns.

The final result shows the users most likely to churn in the future - time for your CSMs to spring into action!

02.

Understanding Churn

This example shows how you can easily understand the drivers of churn.

It's super simple - all you need to do is wrap your PREDICT function with EXPLAIN ! Infer then uses Explainable AI to examine your churn prediction model, and figure out what the drivers of churn are.

The final result shows the importance of each column (features) in your churn model. Now you can understand where your churn is coming from, and you can use this information to action on and improve retention!

03.

Forecasting Revenue Loss

This example shows how you can use your churn model predictions to forecast the expected loss of revenue due to churning customers.

Now simply multiply the contract value ($$$) of your customer with the probability of them churning to forecast revenue loss!

The final result orders the results by largest expected revenue loss, so now you can focus on the most important customers who are at risk of churning! Reach out and save them while you still can.

Explore other use cases

Text Analysis

Customer Segmentation

Lead Scoring

Demand Forecasting

User Similarity Search

We SELECT data from the customer table. These are used as the input to the prediction model function, PREDICT .
The PREDICT function is used to predict a value for the churned column for each row in the customer table, and the ignore parameter is being used to specify that the customer_id column should be ignored in the prediction.
To look at potential future churn, we select the rows where the model predicts churn (prediction='Yes') and customer churn is undecided (churned IS NULL).
The resulting rows are then ordered by the probability column in descending order. We return all of the columns from the customer table (* indicates to return all columns), as well as the prediction and probability columns.

The final result shows the users most likely to churn in the future - time for your CSMs to spring into action!