Toan Hoang

Drawing a Drop Off Sankey Chart in Tableau

Toan Hoang

Tableau Magic has experienced a fantastic launch and I am extremely grateful for your support. One of the ways I have been monitoring the growth of Tableau Magic is via Google Analytics, and via the Audience Cohort Analysis. Cohort Analysis is a subset of behavioural analytics and works by creating subsets of your total and looking at patterns. Google Analytics Audience Cohort Analysis allowed me to see the number of people who returns in subsequent weeks. I am not a huge fan of the visualisation of this data so decided to create a Drop Off Sankey Chart in Tableau. In this tutorial, we will go through the step by step process to create this.

Note: This is an alternative type of data visualisation, and sometimes pushed for by clients. Please always look at best practices for data visualisations before deploying into production.

Data

Load the following data into Tableau Desktop / Public.

DateSegmentPercentagePath
1Members10
1Members1101
2Members0.70
2Members0.7101
2Leavers0.30
2Leavers0.3101
3Members0.50
3Members0.5101
3Leavers0.20
3Leavers0.2101
3Ex Members0.30
3Ex Members0.3101
4Members0.40
4Members0.4101
4Leavers0.10
4Leavers0.1101
4Ex Members0.50
4Ex Members0.5101

Note: we need two records for each Country as we are going to be drawing lines and using densification to get more points on our canvas. For more information, check out our article on Data Densification.

Calculated Fields

With our data set loaded into Tableau, we are going to create the following Calculated Fields, Bins and Parameters:

Create Path (bin)

Create a Parameter called Total Population

Create a Parameter called Sigmoid Size Factor

Now let us create the following Calculated Fields:

Index

INDEX()-1

TC_Percentage

WINDOW_MAX(MAX([Percentage]))

TC_Curve Height

(1 /(1+EXP(-6))* (1/[Sigmoid Size Factor]))

TC_Label

[TC_Percentage]*[Total Population]

TC_Lookup -1

LOOKUP([TC_Percentage],-1)

TC_Lookup -2

LOOKUP([TC_Percentage],-2)

X

((IF [Index] < 51 THEN [Index] ELSE 101-[Index] END)*2*0.12)-6

Y

IF WINDOW_MAX(MAX([Segment])) = "Members" THEN
    IIF([Index] < 51,0,[TC_Percentage])
ELSEIF WINDOW_MAX(MAX([Segment])) = "Leavers" THEN
    IF [Index] < 51 THEN
        1 /(1+EXP(-[X]))* (1/[Sigmoid Size Factor])
    ELSE
        (1 /(1+EXP(-[X]))* (1/[Sigmoid Size Factor]))+([TC_Percentage])
    END
    +[TC_Lookup -1]
ELSE
    IIF([Index] < 51,0,[TC_Percentage])
    +[TC_Lookup -1]+[TC_Curve Height]
    +[TC_Lookup -2]
END

Let us take a moment and look at this Calculation:

So now that we have created a lot of Calculated fields, we will now put this together into a Worksheet.

Worksheet

We will now build our worksheet:

If all goes well, you should now see the following:

Now we can adjust the cosmetics to get the look and feel that we desire:

If all goes well, you should now see the following:

and boom, we are done. We can now see, for a given initial population, what percentage of people (or members) come back in week 2, 3 and 4, and how many people drop off and at which week.

Imagine the following scenarios:

You can find my copy of this Visualisation on Tableau Public at https://public.tableau.com/profile/toan.hoang#!/vizhome/DropOffSankey/DropOffSankey

Summary

I hope you all enjoyed this article as much as I enjoyed writing it and as always do share the love. Do let me know if you experienced any issues recreating this Visualisation, and as always, please leave a comment below or reach out to me on Twitter @Tableau_Magic.

If you like our work, do consider supporting us on Patreon, and for supporting us, we will give you early access to tutorials, exclusive videos, as well as access to current and future courses on Udemy:

Also, do be sure to check out our various courses:

Exit mobile version