Instantly share code, notes, and snippets.

@rooneyroy

rooneyroy / gist:f34276fa3049a598d2401ef64189133b

  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Embed Embed this gist in your website.
  • Share Copy sharable link for this gist.
  • Clone via HTTPS Clone using the web URL.
  • Learn more about clone URLs

Smitan Pradhan

Smitan Pradhan

Data Science Portfolio

  • Melbourne, Australia
  • Custom Social Profile Link

Predicting churn in Telecom Industry - Advanced ML

less than 1 minute read

In the telecom industry, customers are able to choose from multiple service providers and actively switch from one operator to another. In this highly competitive market, the telecommunications industry experiences an average of 15-25% annual churn rate. Given the fact that it costs 5-10 times more to acquire a new customer than to retain an existing one, customer retention has now become even more important than customer acquisition.

For many incumbent operators, retaining high profitable customers is the number one business goal.

Background: To reduce customer churn, telecom companies need to predict which customers are at high risk of churn. We have been hired by a telecom industry giant to look at customer level data and identify customers at high risk of churn and identify the main indicators of churn.

Problem Statement: We need to build a predictive model using advanced Machine Learning algorithms in order to predict the customers at high risk of churn along with the key indicators of churn.

Link to the project code

This case study has been completed with the help of my team mate Koushal Deshpande. Thanks Koushal for your help and your key insights!

You may also enjoy

Forec app - a local solution to a global pandemic.

1 minute read

How can we as Data Scientists help our community during this pandemic?

White box vs Black Box models: Importance of interpretable model in today’s world

6 minute read

In today’s world, more and more focus is on ensuring how black box models can be interpreted and if they are really required

Sedans and Colours - Are certain colours more prevalent in sedans?

3 minute read

Which colour sedans people prefer buying and is there a difference in the trend between different segments of the cars?

Tweet Sentiment Analysis - Naive Bayes Classifier

To classify the sentiments behind a large corpus of tweets

Gramener Logo

Churn Analysis: How to Retain Customers [Telecom Case Study Inside]

This is an article on churn analysis and talks about the importance of customer churn analysis and how it works.

Churn analysis uses insights to help B2B and B2C organizations identify the reasons for customer attrition. Let’s explore churn analysis with simple examples.

Ben is dissatisfied with frequent call connectivity issues on the LoSignal network and decides to terminate his contract with them and switch over to their main competitor. 

AcmeCorp decides that the customer experience offered by Manageo’s Project Management tool is not worth what they’re paying for, and decides to cancel their subscription. 

These are the examples of customer churn and how companies lose clients/customers.

Table of Contents

Customer Churn Rate is the percentage of customers who stopped using a product or service during a particular time frame. 

The first case talks about a B2C situation , where Ben is one of the hundreds of thousands of LoSignal’s customers. In LoSignal’s case, the revenue per customer is comparatively lower, as is the cost of acquiring new customers.

In the second B2B case with Manageo , the customer experience plays a more crucial role in preventing AcmeCorp from leaving, because the number of customers is significantly lower, and the cost of acquiring new customers is higher. The revenue generated by every customer like AcmeCorp is also high.

Did you notice?

In the above cases, Ben and AcmeCorp have “churned”.

In today’s growing marketplace, customer attrition is commonplace. Consumers have a wide variety of options to choose from, each one offering something different — a better customer experience, lower pricing, or better products and services. So, it’s vital for organizations to perform customer churn analysis to retain their customers, be it a B2C or a B2B scenario.

What is Customer Churn Analysis?

Churn analysis helps in determining the trend of customer attrition. The impact of churn analytics efforts are visible in improved customer retention rate.

The right churn analysis insights will help you understand 3 key points through the analytics toolkit:

  • Why are customers churning? What are the key customer dissatisfaction drivers? You can tackle this with descriptive and diagnostic analytics .
  • How do you identify which customers are going to churn in the coming months? You can answer this with predictive analytics .
  • What should you do to minimize churn? Prescriptive analytics holds the key to this.

Before getting into the analytics part, we need to first understand churn rate.

You can calculate the customer churn rate by factoring in the number of customers who left during a particular time period and the total number of customers at the start of that time period. Here’s the formula for more clarity: 

customer churn rate formula for churn analysis

Next, the goal of customer churn analytics is to determine why did a customer stop using a product or cancel a subscription . This usually involves tracking the customer journey and finding out what actions the customer took just before they quit. 

For instance, did they get stuck performing a certain task? Did they encounter any bugs in the system while submitting a page? Feedback calls, follow-up surveys and questionnaires can also help in identifying reasons for customer churn. 

Customer churn analysis also tries to predict which customers are likely to quit. For instance, a lower number of login sessions indicates low usage, and a high possibility of customers churning. After all, why would they pay for something they don’t use often?

By answering these questions, you can improve your customer experience and retain more clients.

Why is Churn Analysis Important?

The most important factor is that Churn Analytics Results in increasing your profits. Fred Reichheld, the founder of the NPS score system , found that if you retain just 5% of your customers, it results in at least 25% higher profits in the long run. On the other hand, churned customers don’t contribute anything to your revenue.

Check out Gramener’s NPS Analytics solutions , which aim to help you improve the Net Promoter Score of your brand by analyzing customer behavior and sentiments from multiple customer feedback and review sources. Check out our guide on how to calculate NPS with Machine Learning.

Second, retaining customers also means that you will unlock the long-term benefits of customer loyalty. A Bain & Company study found that customers who bought from a particular e-commerce platform for about 3 years, spent about 67% more than customers who had been buying from the platform for about 6 months.

Third, churn analysis reduces costs. Depending on your industry, acquiring new customers is five to 25 times more expensive than retaining an existing one. 

It costs more for businesses to acquire new customers than retain existing ones, because existing customers have already bought into your products or services, the philosophy of your company, and are more familiar with your offerings. 

In fact, customers who have been your customers for quite some time will do the best they can to not churn, because new offerings will mean an unfamiliar way of doing things. 

How Does Churn Analytics Work?

Simply calculating the churn rate is not sufficient to help you decrease it. You need the following data points as well.

Questions to ask your data while performing customer churn analysis

Churn analytics can help you answer the above questions and more. Let’s tackle the “why” question first.  

Why Does Customer Churn Happen?

To answer this, you need to first capture the user behavior before they dropped off and see at which stage they decided to churn. Once you’ve captured the data, you will get a few initial insights.

Here are a few possible scenarios for customer churn in B2B:

  • The overall customer experience is lacking
  • Customers don’t get the value they expected from your product
  • You don’t have a robust onboarding process that helps users start using the offering
  • Your product is too expensive compared to competitors.
  • UI/UX is not intuitive and fluid
  • Bugs and glitches
  • Slow load time or processing time

7 Reasons why customer churn happens

Against each reason, you can start assigning $ values.

For instance, let’s say you’ve found out that 20 of your churned customers, each of whom subscribed to your offering for $100 a month, quit due to poor customer experience. This means that you have lost revenue of 20X$100 = $2000 a month due to delivering a subpar customer experience. 

Once you have determined the reasons for your biggest revenue losses, you can prioritize and work on them.    

Which Customers are Canceling?

As you get more data about the customers who cancel, you can classify them into different buckets based on their demographics, behavior, and the offerings they are using. 

Customer churn is one of the biggest problems of the streaming services industry in the U.S., with 41% of customers churning as of Q1 of 2020 (largely exacerbated by users staying home and wanting to try new services during the Covid-19 lockdowns).  

So you can bucket churned customers based on common characteristics such as: 

  • Whether they signed up for a trial
  • The plan they chose
  • The number of devices they chose
  • Users who experienced a particular update or new feature

This grouping and analysis of customers is called “Cohort Analysis” and it lets you zero in on the common factors that led to customer attrition. 

For instance, a large number of customers who signed up for a particular plan may cancel. Delving into the details, you may find that a competitor has a similar offering at a lower cost. Or a group of customers didn’t convert after the trial because they didn’t find enough value to pay for a subscription. 

You can take targeted action to retain each of your cohorts. While a group of customers responds well to discounts, other customers will stick with you even without them.Some customers love loyalty programs, while for some others, it doesn’t make a significant difference. This clustering of customers helps you understand how you can improve your customer experience.

How Can You Predict Customer Churn?

Once you have a sufficiently large dataset of your churned customers, you will begin to understand which customers are churning, and why. Based on this, you will be in a position to predict which customers will churn next. 

How can you achieve this? You can train a Churn Analysis Machine Learning model based on the data that will predict which customers are likely to churn. 

There are several algorithms like decision trees, random forests, and logistic regression that can be deployed based on your industry, offerings, and other factors. 

Each will give a different accuracy rate. Based on these algorithms, a robust AI model is tailored for our organization.

Let’s look at a real-time case study based on a set of algorithms that we applied for one of our clients, a leading telecommunications provider in the U.S. 

Case Study: Churn Analysis for a Leading Telecom Client

Customer churn in the telecom industry is very common due to huge competition.

Did You Know!

The U.S. telecom industry witnesses an annual churn rate of 30-35% . This means that for every 100 people who start the year as customers, 30-35 opt out by the end of the year.

One of the measures that companies use to predict who will churn, target the right customers and deploy interventions to prevent this. 

gramener case study for delivering churn analytics offering for a leading telecom client

The Challenge

Gramener worked with a major telecom operator to identify customers who are likely to churn so that the firm could target its marketing efforts in the right direction. 

The Approach

Gramener applied a series of classification models based on customer behavior, demographics, and network behavior . The result was a series of churn prediction models of increasing accuracy.

The Outcome and Impact

Through derived variables and hybrid models, Gramener identified savings of over 60% on acquisition cost through targeted marketing. The process of identifying targets was also fully automated by the model.

Gramener's customer churn prediction model to do analysis for b2b customer retention

Reduce Customer Churn with Gramener

Gramener's Customer Experience Analytics Offerings

Customer churn analytics is a great way to find out which customers are canceling, why they are canceling, and who will cancel next. Taking the analytics route will also let you reap the maximum ROI from customer experience (CX) efforts .

It helps you segment them and target marketing campaigns according to their demographics and behavior. It’s vital to provide excellent customer service if you want to keep your current customers. If customer service is inadequate, it does not take long for customers to migrate to another service provider. When your customers rate your services, you’ll be able to see how satisfied or dissatisfied they are. The customer sentiment analysis technique is used to reduce customer attrition.

The entire process helps you retain more customers, build a brand, and get you more repeat customers and referrals which will help you uncover your customers’ pain points and give you an opportunity to address them. This in turn will lead to an improved customer experience.

Are you interested in knowing more about how you can reduce customer churn? Talk to us today.

contact gramener to know how to improve customer retention with analytics

Share with Friends:

About The Author

telecom churn group case study github

Varun Bala Krishnan

1 thought on “churn analysis: how to retain customers [telecom case study inside]”.

' src=

Could you please provide more details about the features. what type of data pre processing, feature engineering, modelling were done at a high level.

Leave a Comment Cancel Reply

Your email address will not be published. Required fields are marked *

telecom churn group case study github

  • More Networks

IMAGES

  1. GitHub

    telecom churn group case study github

  2. GitHub

    telecom churn group case study github

  3. telecom-churn-prediction · GitHub Topics · GitHub

    telecom churn group case study github

  4. Churn Analysis: How to Retain Customers [Telecom Case Study]

    telecom churn group case study github

  5. GitHub

    telecom churn group case study github

  6. Customer Churn Analysis Case Study on Telecom Industry Project

    telecom churn group case study github

VIDEO

  1. Portfolio Project 3

  2. EVM603 GROUP CASE STUDY

  3. WQD7005 Data Mining Group Project: Telco Customer Churn

  4. Predicting Customer Churn

  5. intelligent customer Retention using machine learning enhanced prediction telecom customer churn

  6. Churn Prediction in Telecom Industry using Logistic Regression

COMMENTS

  1. GitHub

    Telecom churn case study where, Based on customer behavior (such as the monthly bill, internet usage, etc.) to predict whether a particular customer will switch to another telecom provider or not (i.e. churn or not). Visualised data and provided some insights on the same. The case study has the following files: Telecom_Churn_FINAL -- python ...

  2. GitHub

    In the telecom industry, customers are able to choose from multiple service providers and actively switch from one operator to another. In this highly competitive market, the telecommunications industry experiences an average of 15-25% annual churn rate.

  3. Telecom Churn Prediction Group Case Study

    To reduce customer churn, telecom companies need to predict which customers are at high risk of churn. In this project, you will analyze customer-level data of a leading telecom firm, build predictive models to identify customers at high risk of churn. In this competition, your goal is to build a machine learning model that is able to predict ...

  4. GitHub

    Telecom-Churn-Case-Study In the telecom industry, customers are able to choose from multiple service providers and actively switch from one operator to another. In this highly competitive market, the telecommunications industry experiences an average of 15-25% annual churn rate.

  5. Supriy-a/Telecom-Churn-Case-study

    Problem Statement To reduce customer churn, telecom companies need to predict which customers are at high risk of churn. The goal is to build a predictive model that can predict the chrun propensity for a customer accurately. Identify key features that indicates the customer behaviour to churn. Approach

  6. KBanerjee90/Telecom-Churn-ML-Group-Case-Study

    Telecom Churn - ML Group Case Study Github. Contribute to KBanerjee90/Telecom-Churn-ML-Group-Case-Study development by creating an account on GitHub.

  7. GitHub

    Contribute to Vishal9628712304/Telecom-Churn-Group-Case-Study development by creating an account on GitHub.

  8. Telecom Churn Case Study · GitHub

    gistfile1.txt. # Load the given files. # Collate the 3 files in a single file. # Understand the structure of the collated file. # Make bar charts to find interesting relationships between variables. # Make Box plots for numeric variables to look for outliers. # Impute the missing values, and perform the outlier treatment.

  9. Predicting churn in Telecom Industry

    In this highly competitive market, the telecommunications industry experiences an average of 15-25% annual churn rate. Given the fact that it costs 5-10 times more to acquire a new customer than to retain an existing one, customer retention has now become even more important than customer acquisition. For many incumbent operators, retaining ...

  10. Telecom Churn Case Study

    Telecom Churn - Case Stud ¶. In [1]: # lets import the required libraries and packages import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns from sklearn.model_selection import GridSearchCV from sklearn.linear_model import LogisticRegression from sklearn import metrics from sklearn.model_selection import ...

  11. GitHub

    Find and fix vulnerabilities Codespaces. Instant dev environments

  12. GitHub

    #Telecom Churn Case Study. Problem Statement. Business problem overview. In the telecom industry, customers are able to choose from multiple service providers and actively switch from one operator to another. In this highly competitive market, the telecommunications industry experiences an average of 15-25% annual churn rate.

  13. Anandpratap09/TELECOM-CHURN-DATA-CASE-STUDY

    Contribute to Anandpratap09/TELECOM-CHURN-DATA-CASE-STUDY development by creating an account on GitHub.

  14. Telecom-Churn-Case-study

    To reduce customer churn, telecom companies need to predict which customers are at high risk of churn. In this project, we analysed customer-level data of a leading telecom firm, build predictive models to identify customers at high risk of churn and identify the main indicators of churn. You can also access the Live web-app here.

  15. Telecom Churn Case Study

    Telecom Churn Case Study - UpGrad Python · telecom_churn. Telecom Churn Case Study - UpGrad. Notebook. Input. Output. Logs. Comments (0) Run. 6.2s. history Version 1 of 1. menu_open. License. This Notebook has been released under the Apache 2.0 open source license. Continue exploring. Input. 1 file. arrow_right_alt. Output. 0 files.

  16. Projects · telecom-churn-group-case-study · GitHub

    GitHub is where people build software. More than 100 million people use GitHub to discover, fork, and contribute to over 420 million projects.

  17. Churn Analysis: How to Retain Customers [Telecom Case Study]

    Let's look at a real-time case study based on a set of algorithms that we applied for one of our clients, a leading telecommunications provider in the U.S. Case Study: Churn Analysis for a Leading Telecom Client. Customer churn in the telecom industry is very common due to huge competition. Did You Know! The U.S. telecom industry witnesses an ...

  18. Dronicho/moscow_city_web

    Contribute to Dronicho/moscow_city_web development by creating an account on GitHub.

  19. Telecom Churn Case Study

    Telecom Churn Case Study Python · telecom_churn. Telecom Churn Case Study. Notebook. Input. Output. Logs. Comments (0) Run. 28681.2s - GPU P100. history Version 3 of 3. License. This Notebook has been released under the Apache 2.0 open source license. Continue exploring. Input. 1 file. arrow_right_alt. Output. 0 files.

  20. Issues · ManasBadu/Telecom-Churn-Group-Case-Study · GitHub

    Telecom Churn Group Case Study. Contribute to ManasBadu/Telecom-Churn-Group-Case-Study development by creating an account on GitHub.

  21. TravelTripperWeb-QASites/courtyard-by-marriott-moscow-city ...

    Contribute to TravelTripperWeb-QASites/courtyard-by-marriott-moscow-city-center-moscow development by creating an account on GitHub.

  22. New ITU case study maps the Moscow 'smart city' journey

    A new ITU case study offers an evaluation of Moscow's progress in meeting the objectives of its 'smart city' strategies and the United Nations Sustainable Development Goals (SDGs). The case study , Implementing ITU-T International Standards to Shape Smart Sustainable Cities: The Case of Moscow, was undertaken using the Key Performance ...

  23. Case Study: Churn Prediction

    Case Study: Churn Prediction 4 minute read Business Problem Overview. In the telecom industry, customers are able to choose from multiple service providers and actively switch from one operator to another. In this highly competitive market, the telecommunications industry experiences an average of 15-25% annual churn rate.

  24. GitHub

    A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior.

  25. GitHub

    Contribute to lokeshDSC43/Telecom-Churn-Case-Study development by creating an account on GitHub.