1 of 27

Telco Customer Churn

2 of 27

Our Team

2

Hilman Hibabullah

Project Lead

Dosmaria Bunga Valensia Hutajulu

Data Cleaning & Analysis

Figra Tama Ilahi

Data Cleaning & Analysis

​

Muhammad Fariz Wadzdi

Visualizatiion & Presentation

3 of 27

Executive Summary

3

Customers with a tenure of 1 to 5 months are the most likely to churn.

​

Our Fiber Optic not well performed. From 3096 customer who choose fiber optic 1297 is Churn

​

There’s many of our customers is seing our compotitors

1869 customers or 26,58% of the telco's 7032 customers have left the company.

We must Improve existing product services to make our customer stay with us

​

We must adjust and make a better signal quality

​

Adjust our pricelist and make sure the exact price for each products

3

4 of 27

Overview

01

​

Problem Statement

02

Exploratory Data Analysis

03

Clustering and Customer Segmentation

04

Recommendations

05

Appendix

06

Table of Content

4

5 of 27

Overview

6 of 27

A company called Telco was founded in 2015.

6

Services

Phone Service

​

6.352 customers

Internet Service

​

5.512 customers

Company Highlights

7032 Customers

3.549 Customers

3483 Customers

There are 1 - 72 tenure.

6

Total Billing Customer

​

$ 16.056.169

7 of 27

Data Overview

The Telco customer churn data contains information about a telco company that provided home phone and Internet services. Dataset contains :

  • Service that each customer has signed up for - phone, multiple lines, internet, online security, online backup, device protection, tech support, and streaming TV and Movies
  • Customer account information - how long they’ve been a customer, contract, payment method, paperless billing, monthly charges tenure, and total charges
  • Demographic info about customers - gender, age range, and if they have partners and dependents.

7

7

8 of 27

Methodology

8

Defining Problem

Exploratory Data Analysis

Analysis

(Cluster Analysis, Customer Segmentation)

Data Preparation

(Data Cleaning, Change Data Type)

Insight & Recommendation

8

9 of 27

Problem Statement

10 of 27

26,58%

1869 customers or 26,58% from 7032 customers was Churn

​

10

What is our problem?

11 of 27

How to reduce churn rate to 10% at the end of next year?

Find the characteristics of our Customer, then retain them with our good quality services and make some public campaign

11

11

12 of 27

Root Cause Analysis

12

12

13 of 27

Exploratory Data Analysis

14 of 27

Most customer who churn are customers with tenure 1-5 month (New Customer)

14

14

  • Tenure 1 months is the highest with 380 customers.
  • On overall, 1-5 as we call new customer is the most highest to churn with 744 customers. Which is almost 39% for customer churn
  • We can describe the customer with::

<= 5 Month = New Customer

> 5 & <=12 Month = Regular Customer

> 12 <= 24 Month = Prospect Customer

> 24 Month = Loyal Customer

​

  • However there are still customer with 72 tenure is churn

15 of 27

Total Customer of Internet service users is 3,096 and 1,297 customers is churn

15

16

  • Fiber Optic is the most Internet Service that customer pick, but The most customer churn in Internet Service is from Fiber optic.
  • Fiber Optic is tend to have a bad quality for customer experience.
  • Total DSL is 2,416 and 459 is churn.
  • Total no using Internet Service is 1,520 and 113 is churn.

16 of 27

There are some Customers with < 10 tenure have a high MonthlyCharges

16

16

  • There some customer with only 1 tenure but have MonthlyCharges almost $110, but after that the customer was churn
  • It’s mean that there’s some new customer who just want to try all our product
  • Therefor there’s customer with 701 tenure but the MonthlyCharges is only around $20, it means that this customer only have 1 of our product for several years

​

17 of 27

Each telco company's product has a wide variety of prices.

17

  • Internet services are generally costly. The most expensive option is fiber optic, with an average cost of 50.01, while DSL costs 24.9.
  • The cost of Phone Service, with a minimum of 18.25 a maximum of 21.45, and a mean of 19.95.
  • The typical cost of multiple lines, Online Security, Online Backup, and Tech Support is approximately 5.
  • The typical cost of Streaming TV, and Streaming Movies is approximately 10.

Service

Price

Min

Max

Mean

Phone Service

18.25

21.45

19.95

DSL

23.45

26.05

24.9

Fiber Optic

47.8

52.3

50.1

Multiple Line

3

7

5.01

Online Security

3.6

6.15

5.04

Online Backup

3.5

6.75

5.14

Device Protection

4.2

5.5

4.8

Tech Support

3.7

6.3

5.04

Streaming TV

8.5

11.5

10.3

Streaming Movies

9

11.6

10.1

17

18 of 27

Clustering & Customer Segmentation

19 of 27

Cluster Analysis

19

We determine amount of cluster by used the highest silhouette score for n cluster

19

20 of 27

Characteristics of each cluster

20

Recommendation

  • Bundling Items (Cross Selling)
  • Notification of new related products.
  • Upselling to higher-priced products.

This cluster contains customers who want to test our services, or can be said new customers, who use telephone services, but are not very interested in Internet Services. So other services such as online backups to streaming movies do not use these services. This causes this cluster not to make a large contribution to the company's revenue

SILVER

Recommendation

  • Send a special appreciation or gift.
  • If there are any new services, make them an exclusive offer.

A loyal customer because the majority of customers in this cluster have a tenure of more than two years, and this cluster has all of the service, and as a result, this cluster is the largest contributor to the company by spending a large amount of money in the company based on the total charge

GOLD

20

21 of 27

Insight & Recommendation

22 of 27

We Have Some Sugestion Recommendation for Company

22

1-5 tenure is our problem because customer churn is high.

We recommend the company to eliminate the option to sign up for a one-month contract then give a new customer special offer for try for free our services for 3 days

Expected Result :

​

Customers with a tenure of 1-5 months decreased by arround 10%

Fiber optic is the most expensive product ($50), and customers must also have phone service if they want fiber optic. Maybe the customer is dissatisfied with the quality of fiber optic for which they have already paid a high fee.

We need to improve fiber optic quality while also recommending price reductions.

Expected Result :

​

Reduce fiber optic churn and retaining existing customers

Each Telco Company’s product has a wide variety of prieces

We recommend fixing the price with a fix number

Expected Result :

​

So that customers get the best possible price and the churn rate is reduced by about 3%

22

23 of 27

General Recommendation

23

We want to reduce churn rate

We recommend the company to build a strong predictive model to anticipate customer to churn

Expected Result :

Company know which new customer have tend to churn when new customer sign for new contract. So Company can approach another option for them

Our services is still cant dominate the market

We recommend company must create a campaign that is sustainable for longlast time

Expected Result :

​

Our Brand can well known in public, so we can deep dive more into the market

23

24 of 27

“...maintaining existing customers is a more important business issue than acquiring new customers.”

​

​

Hung, . Chihli, Tsai and Chih-Fong, Vol. 34, 2008, pp. 780– 787.

24

25 of 27

Thankyou

26 of 27

Appendix

You can find the dataset on Kaggle here

You can find Python code Analysis here

You can find Tableau Dashboard here

27 of 27