Case study – Understanding customer churn

For the group project, you will work with a fictitious dataset from Databel, a mobile network
provider that wants to decrease its churn rates (i.e. customers leaving the company). You will
have to apply the knowledge that you gained during the module’s seminars on analysing data
and creating dashboards on Tableau. You will have to complete the project on Tableau
Public (instead of the training platform), so you can share your dashboard with the examiner.
You can download the platform from here:
https://www.tableau.com/products/public/download
More specifically, as a group, you will have to complete the following:
 Explore a dataset and create calculated fields in Tableau to get a better understanding
of why customers are churning.
 Investigate why customers are leaving Databel.
 Create Tableau interactive dashboards and arrange them into stories, so that you can
share your results with stakeholders.
The churn rate in this case study is defined as: Churn rate = customers lost/total number of
customers
Remember to check the document with the Metadata to understand what each column in the
dataset means.
The tasks that you need to complete are described below. Use the hints if you are not sure
how to complete the tasks. The group will have to make decisions about the format that will
be used in the graphs
Task 1: Create two measures to check if the count of customer ids is equal to the count of
unique customer ids. This is to ensure that there are no duplicate rows in the dataset.
Hint: Load the CSV file Databel – Data.csv from the folder you have stored the downloaded
file. To do that, open an empty Tableau workbook if you don’t have one already open. On the
Tableau start page, you need to click “Text file” on the left menu. Find the folder in which
you have stored the dataset and choose the CSV file.
Open the metadata information file to find the fields required for the task. Create two
calculated fields and add them to a new sheet:
Number of Customers = COUNT ([Customer ID])
Number of Unique Customers = COUNTD ([Customer ID])
Add the two new measures to an empty sheet and name it Datacheck.

Task 2: Use an IF statement to convert the Churn Label column into a Churned column. It
should contain a 1 if a customer churned, and 0 in case a customer didn’t churn.
Hint: Create a new calculated field named ‘Churned’ and use the following formula:
IF [Churn Label] = ‘No’ THEN 0 ELSE 1 END
Task 3: Create a calculated field Number of Churned Customers using the previously created
Churned column to do so.
Hint: Use the following formula:
SUM ([Churned])
Task 4: Calculate the Churn Rate as a percentage.
Hint: Create a new calculated field named “Churn Rate” and use the following formula (the
Number of Customers can also be used as the measures match the number of unique
customers):
[Number of Churned Customers]/[Number of Unique Customers]
You can format the Churn Rate as a percentage by right clicking and going to ‘Number
Format’ in ‘Default Properties’.
If you add the new field to a new sheet the churn rate should be 26.86%
Task 5: Visualise the reasons why customers churn using Number of Customers and Churn
Reason in a Column Chart.
Hint: Create a new sheet and name it accordingly. You can create a column chart by adding
Number of Customers to Columns and Churn Reason to Rows. You can filter the churners
using the Churn Label.
Task 6: Order the churn reason descending do the most popular churn reason appears on top.
Calculate Number of Customers as ‘Percent of Total’.
Hint: Click on the Sort button in the graph or right click on the Churn Reason and select Sort.
You can change the Number of Customers to a percentage of calculation by right clicking on
the field in the Columns and using a Quick Table Calculation.
Task 7: Create a visualisation of your choice to display the churners by churn category in a
new sheet. You can use the Show me feature on the top right to create a tree map, a highlight
table, a pie chart or something else. Make sure to include the ‘Null’ category, as there will be
churners that didn’t include a churn reason.
Hint: Create a new sheet and name it accordingly. You can filter for churners using the Churn
Label.
Task 8: Use a percent of total calculation, so it’s easy to spot the percentage of churners it
represents.
Hint: You can add a ‘Quick Table Calculation’ by right-clicking on the number of customers
measure. Depending on the type of graph, it can be found in the Marks, Column, or Row
section. For example, in the case of a Tree Map graph, it can be found in the Marks section.
Task 9: Create a new sheet and build a map. You can find more details on what should be
included in the next step.
Hint: You can create a map by double clicking on the State field. You can play around with
the Color and Size marks card to create different visualisations.
Task 10: Make sure to add Churn Rate, Number of Customers, and Number of Churned
Customers to Tooltip. This will be very useful for your stakeholders when you publish the
map on a dashboard. Make use of Size or Color so it’s easy to spot states with a higher churn
rate.
Hint: Drag the respective measures to Tooltip in the Marks card to add them. States with high
churn rates can be shown easily by e.g. dragging Churn Rate to Color.
Task 11: Check the churn rates for the following demographic dimensions: gender, senior and
under 30. You can do that by adding the demographics dimensions to new sheets (use one
sheet per dimension) and drag the Churn Rate to the Text in the Marks area. Do you notice
any significant differences among demographic groups?
Hint: Check the churn rate for senior citizens and compare it to the average. You will notice
that the churn rate for senior citizens is around 10% higher than the average.
Task 12: Based on the above insight, it would make sense to analyse age in general. You will
create different age bins and a combo chart visualising the number of customers per bracket
and their respective churn rates. Start by creating bins for Age. You can use any bin size you
prefer, but it is often a good idea to use a bin size that is easily understandable (e.g. 5 years).
Hint: Right-click Age and select “Create”, “ Bins…”. A bin size of 5 would fit well in this
case.
Task 13: Create a dual-axis chart that shows the number of customers and churn rate for
every age bracket in a new sheet.
Hint: You can start by adding Age (bin) to Columns, and Number of Customers and Churn
Rate to Rows. Then, you can right-click on any of the two measures in the Rows section and
select “Dual Axis”.
Task 14: Visualise the number of customers as an area or a bar chart; you can keep the
visualisation you like the most. Format the churn rate as a line and select any colours you like
to make the graph visually appealing.
Hint: You can change the chart format in the Marks card to “Area”, “Bar” or “Line”.
Taske 15: Databel offers group contracts to customers from the same household. The
advantage for the customer is a discounted rate, while it is a great way for Databel to grow its
customer base. Your task is to analyse if customers that are part of a group have indeed a
lower phone bill, and if it has an impact on the churn rate. Start the analysis is a new sheet by
changing the currency of Monthly Charge to $ in the Data Pane. Then, create a simple bar
chart that plots the average Monthly Charge by Number of Customers in Group.
Hint: Right- click on the Monthly Charge in the Data Pane and select Default Properties and
then Number Format…Select Currency (Custom) and make sure that the $ is included in the
Prefix.
Create a new sheet and give it a relevant name. Creating a bar chart using the Number of
Customers in Group measure does not give the desired results. Convert it to a dimension so
you can easily compare the average Monthly Charge rate.
Task 16: It looks like that the Monthly Charge is significantly lower for people who are in a
group of 2 or more. Strengthen the message by adding Group to the chart. Try adding it to
different fields, and choose the one you feel most comfortable with.
Hint: Check how adding Group in the Columns, Rows or in the Marks area changes the
graph. Choose the version that you feel communicates the message more clearly.
Task 17: Add the churn rate to the right axis and convert it to a line.
Hint: Drag Churn Rate to the right until a dotted line appears and release to create a dual axis
graph. You may want to format the colours to make the information clearer.
Task 18: The “0” on the graph is a bit confusing, as it refers to customers that are not in a
group. Change alias to “Not in a group”. Note that this will work only if Number of
Customers in Group is set as a Dimension.
Hint: You can edit the alias of a field by right-clicking on it in the Data pane and selecting
Aliases…
Task 19: The graph can be further improved by adding a dynamic parameter that allows the
stakeholders to interact with the sheet and investigate the number of customers per group, the
number of churned customers per group etc. Create a new parameter, call it Pick Metric,
change the Data type to “String” and change Allowable values to List. Type “Avg Monthly
Charge”, “Number of Churned Customers” and “Avg Customer Service Calls” inside the list.
Hint: In the Data pane, click the drop-down arrow in the upper right corner and select “Create
Parameter”.
Task 20: Create a new calculated field Metric that uses a CASE statement to select a different
measure based on the parameter. It should be structured like this:
CASE [Pick Metric]
WHEN “Number of Customers” THEN [Number of Customers]
WHEN “Avg Monthly Charge” THEN AVG([Monthly Charge])
WHEN “Number of Churned Customers” THEN [Number of Churned Customers]
WHEN “Avg Customer Service Calls” THEN AVG([Customer Service Calls])
END
Note you should add the average Monthly Charge and Customer Service Calls , so there is
fair comparison between the groups.
Hint: You cannot combine aggregate and non-aggregate measures. Number of Customers and
Number of Churned Customers are already aggregated but Customer Service Calls and
Monthly Charge are not. You need to wrap both in an AVG() function.
Task 21: Replace Monthly Charge from the visualisation from the previous task with your
new field Metric and show the parameter. Your graph should now adapt dynamically.
Hint: Your calculated fields are case sensitive so if it doesn’t work straight from the get go
you might have made a spelling error.
Task 22: Next, you are going to explore the influence of the Unlimited Data Plan on churn
rate. Databel has a hypothesis that people who are not on an unlimited plan are more likely to
churn. To test the hypothesis, start the analysis by creating a quick text table that displays the
churn rate for customers who have the unlimited plan and for customers who don’t.
Hint: Double click Unlimited Data Plan and Churn rate to add them to the sheet.
Task 23: Based on the previous text table, it seems that the opposite is true; i.e. customers
who are on an unlimited plan are more likely to churn. It would be useful to have an idea of
how much mobile data in gigabyte they are using on a monthly basis. Create three buckets in
a new calculated field Grouped Monthly GB Download: customers who consume less than
5GB monthly, between 5 and 10 GB, or 10 or more GB.
Hint: You can use an if statement to create the calculated field. Your field should look like
this:
IF [Avg Monthly GB Download] < 5 THEN “Less than 5 GB”
ELSEIF ____ < ___ ____ ____
ELSE ____
END

Task 24: Create a bar chart visualising churn rate using your newly created calculated field
and the Unlimited Data Plan dimension. The goal of the chart is to compare the churn rate in
these different buckets for people who are an unlimited plan vs people who are not.
Hint: You can create a Bar chart by adding your new calculated field and Unlimited Data
Plan to Columns, and Churn rate to Rows. You can also try adding one of these dimensions
to Color to strengthen your message.
Task 25: The analysis requirement given by Databel includes a request to analyse the
international activity of customers and its relationship to churn. They are curious about the
behaviour of customers who call internationally, and/or if paying for an international plan
influences their loyalty. Create a text table that shows the churn rate for the different
international dimensions from the metadata sheet. The columns should show if a customer
has an international plan or not, and the rows should show if the customer is internationally
active or not.
Hint: You can create a text table by double-clicking the three respective fields in the Data
pane. You’ll want to use the Intl Plan, Intl Active and Churn Rate.
Task 26: Convert your table to a “highlight table”. Edit the graph so it displays a high churn
rate with a red background, and a low churn rate with a green background. Feel free to add
any information to Tooltip you deem relevant.
Hint: You can use Show me to convert a text table to a highlight table. You can edit the Color
in the Marks card. You can select the “Red-Green Diverging” color palette and tick the
“Reversed” box to make a high churn rate appear in red.
Task 27: Before you start compiling your analysis in a dashboard, it is important to look at
the remaining factors that might affect customer churn rate, namely payment method, the
contract type, and how many months a person is a customer at Databel. Create a scatter plot
using the average account length and the churn rate. Secondly, you can add information about
the contract and how customers pay to the scatter plot.
Hint: You can create a scatter plot by double-clicking Account Length (in months) and Churn
Rate. Afterwards you can add Contract Type and Payment Method to the scatter plot in
different ways, but try to create a graph that conveys a key insight in the easiest way possible.
Task 28: It appears that customers who are on Month-to-Month contract are way more likely
to churn. Put extra focus on this message by changing the size of the dots using the
appropriate measure.
Hint: There are multiple metrics you can add to size, but Number of Churned Customers is
really perfect here. While doing this you can easily spot which category contains the most
churners.
Task 29: Next, you will start building your dashboards. The first dashboard should be an
overview of your analysis, and should contain key performance indicators (KPIs), such as the
number of customers and the churn rate. It’s also a good idea to add churn reasons to the
graph, because it explains why customers left Databel.
Create a new dashboard and call it Overview. If you feel the dashboard is rather small, you
can change the size to “Desktop Browser 1000x800” or any other size you see fit.
Hint: You can create a dashboard by clicking the “New Dashboard” button in the bottom
right. You can then double-click on Dashboard 1 to rename it. The Size option can be found
on the left.
Task 30: Add the two main Key Performance Indicators (KPIs) – churn rate and number of
customers – to the dashboard. The KPIs should be easily readable and appear on the top left
of the dashboard. You can reformat your old sheets to do this, or create new sheets to add to
the dashboard. Feel free to add any other KPIs you deem relevant.
Hint: You can change the font size of the KPIs using the “Text” option in the Marks card. It’s
recommended you change the font size to 20 or more. Sheets can be added to a dashboard by
dragging them from the Sheets section.
Task 31: Add the column chart with churn reasons to the space below your KPIs. Add the
Churners by category visualisation to the top right, and the map to the bottom right. You
created all of these graphs in the first tasks.
Hint: You can drag the respective sheet to the place on the dashboard. If you are struggling in
getting it to the correct place, you can use “Horizontal” and “Vertical” from the Objects
section.
Task 32: The contract type has a huge impact on the churn rate of customers, so add this to
the dashboard. You can choose to add a filter or a visualisation that functions as a filter.
Make sure the filter is applied to all graphs on the dashboard.
Hint: You need to add a filter to a worksheet visible on the dashboard before you can add it to
the actual dashboard. Once added you should click “Apply to Worksheets” in the dropdown
menu so it can be applied to all graphs. Alternatively, you can create and add a visualisation
with Contract type to the dashboard and click “Use as Filter”.
Task 33: Enable the churn category visualisation you added in the previous task as a filter.
Display the dashboard title and format the dashboard to your taste to make it visually
appealing. Finally, feel free to add text instructions to the dashboard if you think it helps your
stakeholders.
Hint: You can click the “Show dashboard title” checkbox in the bottom left to display the
title. You can add a text instruction by using “Text” in the Objects section in the bottom left.
Task 34: Create a second dashboard that portrays insights about the age buckets and groups.
You’ll use the dynamic parameter you created in the previous tasks (i.e. 19-21), in
combination with filters to make the dashboard interactive. The dynamic parameter already
works for your group visualisation, so you’ll also apply it to the age buckets visualisation
before adding both to a dashboard.
Make your age buckets visualisation dynamic, so the left axis changes based on the parameter
selection. Do not alter the churn rate that is visible on the right axis.
Hint: You created a Pick Metric parameter you are using in a Metric calculated field. Your
group visualisation is dynamic based on that parameter, but your age visualisation in not yet.
You can make the age visualisation dynamic by replacing Number of customers with Metric.
Task 35: Create a slicer filter for the account length. That way stakeholders can e.g.
investigate how the churn rate is for customers who are with Databel for less than a year. Add
any other filters you deem relevant to the sheet and apply them to all worksheets using this
Datasource.
Hint: You can create a slicer filter by adding Account Length (in months) to filters and
selecting Next. Navigate to the more options arrow on a filter and go to “Apply to
Worksheets” and select “All Using This Datasource”. Dimensions like Contract Type,
Payment Method, State and many others can all reveal interesting insights.
Task 36: Add the age and group visualisation to a new dashboard. Display the parameter and
the other important filters below the title. Reformat your filters so they don’t take up too
much space.
Hint: Filters like Contract Type and Payment Method can reveal quite a few insights. It’s a
good practice to only add the most relevant filters to a dashboard. If you e.g. display 10+
filters, it becomes difficult for the stakeholders to use the dashboard. You created the age and
group visualisations during the previous tasks.
Task 37: Test the parameter and different filters to make sure they apply to both graphs.
Finish by cleaning the dashboard so it looks nice. You could, for example, make the legends
floating to give the graphs more space.
Hint: You can click the more options arrow and navigate to “Apply to Worksheets” to see
which worksheets are influenced by the filter.
Task 38: Your next task is to create a dashboard with the scatter plot about the contract and
payment type and combine it with two KPI cards about customer service calls. Create a new
calculated field that calculates service calls per customer. Format it to have two decimal
points.
Hint: You need to divide SUM([Customer Service Calls]) by the total number of customers to
calculate the average. You can right-click a measure and go to “Default Properties” to change
the number format.
Task 39: Create two new sheets, one with the total customer service calls, and one with the
avg customer service calls. Make sure the text is big enough so you can add it as a KPI to the
dashboard later.
Hint: You can drag Customer Service Calls to text to display it in a new sheet. You can
format it by going to “Text” in the Marks card.
Task 40: Add the scatter plot from the previous task (i.e. Task 27) to a new dashboard. Feel
free to add any filters you like to make the dashboard interactive and place them at the top.
Hint: You can only add filters to the dashboard that are added to the scatter plot visualisation.
Churn Category is an interesting filter to add here, because it will place all marks on 100%
churn rate when you e.g. select “Competitor”.
Task 41: Add the two new sheets about customer service calls to the dashboard and make
them floating. Place them in a space so they don’t overlap with the visualisation. Enable
scatter plot to function as a filter.
Hint: You can make a text card “Floating” by using the dropdown menu.
Task 42: The final dashboard should cover the insights about the Data and international plan.
Create a new dashboard and add the bar chart about the data plan, and the highlight table
about the international plan. Make sure you also include information about the average extra
data and international charges per customer on the dashboard.
Hint: You can calculate the averages charges by dividing the total charges by the number of
customers. You are free to choose how to add them to the dashboard. You could create a new
visualisation or display them as a text.
Task 43: Feel free to add whatever you want without cluttering the dashboard. You can
change dashboard shading, title shading or anything else to make it look pretty. It’s a good
practice to be consistent among dashboards, so feel free to adapt one of your previous
dashboards, too.
Task 44: Combine your four dashboards in a story so it is ready to be published. Create a new
story and add the four dashboards to it. Make sure the overview dashboard is displayed as the
first one.
Hint: You can find the “New Story” button at the bottom right, next to the “New Dashboard”
button.
Task 45: Give the story and the different story points a fitting title. Feel free to change the
formatting as you want.
Hint: There is a “Show title” checkbox on the left, which you can select to show the title. You
can double click on a story point to rename it.
Task 46: Make sure that you save your work, as this is the story you will have to share. Keep
the story hidden from your Tableau Public profile (make sure that the “Show Viz on Profile”
is off”), so you can share it only with the examiners. When you are ready to submit your
work on Blackboard, click on the share button (in the top right area, second button) and click
on Copy link. If you have downloaded Tableau Public on your computer and you work
locally on the project, you need to click first on the File  Save to Tableau Public As…, so
you can upload your work on the cloud before you are able to share it.