Creating NPS Gauges in Tableau

Update April 18, 2020: This has become one of my most popular blogs since I first wrote it back in 2017. Unfortunately, I’ve found that there were certain aspects of the how-to process that I failed to explain thoroughly. I’ve also since discovered an easier technique for building this chart. So, to address both of these items, I’ve added an updated section at the bottom of the blog. Feel free to read through the whole blog as it explains NPS and the basic concept used to create the chart, but if you just need to know how to apply it, I’d recommend skipping to the A New Technique section.

This weekend, Rajeev Pandey, a former Tableau Featured Author and regular in the Tableau community, posted a question on the Tableau Community forum asking if anyone had ever created a gauge to show Net Promoter Score. Here’s the image he shared.

He also reached out to me directly to see if I could help, so I figured why not give it a go.  

What is NPS?

Before jumping into how to create this gauge in Tableau, let’s start with an explanation of NPS. According to Medallia.com, “The Net Promoter Score is an index ranging from -100 to 100 that measures the willingness of customers to recommend a company’s products or services to others. It is used as a proxy for gauging the customer’s overall satisfaction with a company’s product or service and the customer’s loyalty to the brand.” I won’t go into detail about how NPS is calculated, but if you’d like to learn more, you can check out the Medallia’s full explanation.

While there are a number of different ways to visualize this metric, one of the most common seems to be a sort of gauge as seen in the image above.

The Chart

Let’s start with some observations about this chart.

• First of all, it looks basically like a half of a donut chart (and donut charts in Tableau are essentially pie charts with a hole in the middle).
• The first slice of a pie chart in Tableau always starts at 0°, but this chart would need to start at 270°.
• The values of the chart will need to go from -100, starting at 270°, to 100, ending at 90°.

So, how to create it? The biggest challenge of this chart is the fact that it needs to start at 270°. I decided that my approach would be to create a donut chart with 5 pre-defined slices, as shown below.

• Slice 1 – Variable sized slice starting at 0° and ending between 0° and 90°. This slice will be used to show the second part of the NPS (if the value is greater than 0). Otherwise, this slice will have 0 size, making it invisible.
• Slice 2 – Variable sized slice starting and ending somewhere between 0° and 90°, after slice 1. If NPS is less than 100, then this slice will be visible and will appear in a light grey color.
• Slice 3 – Hidden slice starting at 90° and ending at 270°. This slice will always be 180° and will always be the same color as the background, rendering it invisible, essentially making it a half-donut.
• Slice 4 – Variable sized slice starting at 270° and ending between 270° and 360°. This slice will be used to show the first part of the NPS (values between -100 and 0).
• Slice 5 – Variable sized slice starting and ending somewhere between 270° and 360°, after slice 4. If NPS is less than 0, then this slice will be visible and will appear in a light grey color, like Slice 2.

Using this relatively simple idea, I created a very small data set containing five records, one for each slice, with the following information about each:

• Slice # – Numeric identifier of the slice (1, 2, etc.)
• Slice – Name of the slice (Slice 1, Slice 2, etc.)
• Details – Description of the slice and how it’s used.

Strictly speaking, we could get by with just Slice # in the data set, but I decided to add a bit more information to make it less confusing to someone using the template.

Notice that there are no measures in the data set. We’ll create the measure within Tableau, based on the NPS score. Note: I’m assuming that the user will have an NPS score which can be fed to the visualization. For demonstration purposes, I’ll just create a parameter allowing the user to select an NPS.

The Calculations

Calculating the pie chart size measure is a matter of some relatively simple math. To simplify my formulas, I created five calculated measures, one for each slice. Here are the formulas for each.

Value_Slice_1:  IF [NPS]>0 THEN [NPS] ELSE 0 END

Value_Slice_2:  IF [NPS]>0 THEN 100-[NPS] ELSE 100 END

Value_Slice_3:  200

Value_Slice_4:  IF [NPS]<0 THEN 100+[NPS] ELSE 100 END

Value_Slice_5:  IF [NPS]>0 THEN 0 ELSE 100-[Value_Slice_4] END

Note: “NPS” is the name of the parameter noted earlier.

Then, I combined these values into a single measure, which I called Chart Value:

CASE [Slice #]

WHEN 1 THEN [Value_Slice_1]

WHEN 2 THEN [Value_Slice_2]

WHEN 3 THEN [Value_Slice_3]

WHEN 4 THEN [Value_Slice_4]

WHEN 5 THEN [Value_Slice_5]

END

Finally, I created a calculated field called Color:

IF [Slice #]=3 then "None"

ELSEIF [Slice #]=5 OR [Slice #]=2 THEN "Grey"

ELSEIF [NPS]>[Highest Normal Value] THEN "Green"

ELSEIF [NPS]>[Highest Bad Value] THEN "Yellow"

ELSE "Red"

END

Note: This formula references two other calculated fields, Highest Normal Value and Highest Bad Value, which refer to the maximum “Yellow” NPS score and the maximum  “Red” NPS score respectively. In this case, I’ve used values 60 and 20 per Rajeev’s requirement.

Next, based on these calculations, I created a very simple pie chart, as shown below:

I then changed to a dual-axis chart, with the second axis showing a manually sized white circle.

Note: I’ve colored the NPS score the same color as the gauge. Doing this required that I create three separate calculated fields as follows:

Value_Bad:  IF [NPS]<=[Highest Bad Value] then [NPS] END

Value_Normal:  IF [NPS]>[Highest Bad Value] and [NPS]<=[Highest Normal Value] then [NPS] END

Value_Good:  IF [NPS]>[Highest Normal Value] then [NPS] END

I then combined those into the Label as follows: <Value_Bad><Value_Good><Value_Normal>. As only one of these calculated fields can have a value, you’ll only ever see one color.

With that complete, I had the basic NPS chart in place.

But, I was missing one key component. The original chart had a small key right below the gauge showing where the values would change from red to yellow to green.

So, how do we create this? I experimented with a few different options, but all seemed to require use of a dual-axis and, since I was using the dual axis for the hole in my donut, I could not find a way to make it work. In the end, I decided that the best approach would be a custom image with a transparent background, which would float over the pie on a dashboard. While I’d definitely prefer not to having to piece together floating elements to make the chart work, I just couldn’t find any other way (If anyone reading this has another idea, I’d love to hear it.)

Since I would be floating an image on top of the chart, I decided to just use the image to double as my donut hole. The image I ended up creating (using mostly PowerPoint and a little Paint.NET) looks like this (the checkered background indicates transparency).

With this element acting as the donut hole, I removed my dual axis and the center circle, taking it back to the original pie chart. 

From here, it was just a matter of adding the pie to a dashboard and floating the image over top of it. Finally, I added one more sheet to display the actual NPS value, which was then floated over top of the image. 

Here’s the final result:

Required Files

If you have a need to create a gauge like this, then feel free to download my examples from Tableau Public. You can also find the Excel template image here. If your customer requires different ranges for bad, normal, and good, then that will require some slight changes to the image. In addition, if you need to change colors (yes, I know that traffic light colors aren’t necessarily the best option), then you can fairly easily recolor the image.

Update 09/05/2017: New Approach!
The other day, I saw the following comment on this blog post, from Sr. Tableau Developer, Tim Hughes (check out his Tableau Public Profile):

Very cool, Ken!

Instead of floating the key image over the donut, I think it would work to instead

1. Make the key image a full circle with the bottom half white
2. Save the image as a custom shape in your Tableau repository shapes folder
3. Change your second mark card from a white circle to that custom shape (keeping the existing label)

Thanks,
Tim

Okay, so this is a great idea. Essentially, he's suggesting that we dispense with the floating image, go back to our original dual-axis approach and simply replace the white circle with a custom shape. So, I created the image as he suggested. Here's how it looks (again, the checkered area indicates transparency):

As Tim suggested, I created this as a custom shape in my Tableau repository. I then went back to the original NPS chart and changed the white circle to use this custom shape. And here's the result:

Obviously, it looks exactly like my previous version. However, this is a much more elegant solution as it does not require you to float an image over top of the chart. Thanks a lot for the tip Tim!! If you'd like to try this method out, you can find the full circle image here.

Update 09/27/2017: Some Alternatives to NPS Gauges
Since writing this blog post, something has been nagging at me—I’m just not sure that this is really the best way to visualize this data...Read more here: Alternatives to NPS Gauges

Update March 15, 2019: A common problem with this blog has been an inability to make it work with your own data instead of the parameter. I apologize for that as I failed to properly explain how to do it. If you are struggling with this, I'd refer you to the following post on the Tableau Community Forum in which I show how to do this. Note: The post deals with percentage gauges but the process should be almost exactly the same for NPS gauges: Percentage Gauges with Your Own Data

A New Technique (April 18, 2020)

The technique used previously required you to cross-join your data to my data set of five rows in order to create the five pie slices. Fortunately, I’ve since found that there is a much easier way to build this chart, which I’ll share now. The same basic concepts exist so if you haven’t read the full blog, that may be a worthwhile first step.

We’ll start with a sample data file that has 10 customers with each of their NPS scores.

Unlike my previous method, this is the only data we need—we do not need the additional slice data set.

We’ll connect to the data, then create four parameters:

Highest Bad Values – Integer or Float showing the highest value that is considered bad (red color). I’ve set this to 20 in my workbook.

Highest Normal Values – Integer or Float showing the highest value that is considered normal (yellow color). I’ve set this to 60 in my workbook.

NPS Test – NPS Test value. We’ll use this to make sure all the slice colors are set properly before using the real NPS value.

NPS Value – Parameter that will determine whether we use the actual NPS value from the data or the test value. The parameter should be a string and contain two values, “Test Value” and “Real Value”

Now we’ll create the following calculated fields (feel free to just copy these from my workbook):

NPS for Chart

// Which NPS value to use

IF [NPS Value]="Test Value" THEN

    // Use parameter

    [NPS Test]

ELSE

    // Use real value

    [NPS]

END

Slice 1

// Slice 1 size

// Slice starting at 0° and ending between 0° and 90°.

IF [NPS for Chart]>0 THEN

    [NPS for Chart]

ELSE

    0

END

Slice 2

// Slice 2 size

// Slice starting and ending somewhere between 0° and 90°, after slice 1.

IF [NPS for Chart]>0 THEN

    100-[NPS for Chart]

ELSE

    100

END

Slice 3

// Slice 3 size - Will always be 200

// Hidden slice starting at 90° and ending at 270°.

200

Slice 4

// Slice 4 size

// Slice starting at 270° and ending between 270° and 360°.

IF [NPS for Chart]<0 THEN

    100+[NPS for Chart]

ELSE

    100

END

Slice 5

// Slice 4 size

// Slice starting and ending somewhere between 270° and 360°, after slice 4.

IF [NPS for Chart]>0 THEN

    0

ELSE

    100-[Slice 4]

END

Color

// Color for the gauge.

IF [NPS for Chart]>[Highest Normal Value] THEN

    "Green"

ELSEIF [NPS for Chart]>[Highest Bad Value] THEN

    "Yellow"

ELSE

    "Red"

END

Value Bad

// Text to show if the chart's value is Bad.

// Separate text so that we can color differently.

IF [NPS for Chart]<=[Highest Bad Value] THEN

    [NPS for Chart]

END

Value Good

// Text to show if the chart's value is Bad.

// Separate text so that we can color differently.

IF [NPS for Chart]>[Highest Normal Value] THEN

    [NPS for Chart]

END

Value Normal

// Text to show if the chart's value is Bad.

// Separate text so that we can color differently.

IF [NPS for Chart]>[Highest Bad Value] and [NPS for Chart]<=[Highest Normal Value] THEN

    [NPS for Chart]

END

Now that we have all of our calculated fields, we’ll create a “fake” axis on the rows shelf. Just type 0 into the shelf, then right-click the pill and change it to a dimension. Then drag Measure Names to the color card and Measure Values to the angle card. This will put all of your measures in the Measure Values pane—remove everything except the five slice measures.

Now move the Color field to the detail card. Once it’s there, click the icon to the left of the pill and change it to color. Our slices will now be colored by a combination of the measure name and the color.

Now we can set our colors. Set slice 3 to white (or whatever background color you’re using), slices 2 and 5 to a shade of grey, and slices 1 and 4 to the color specified in the Color field (in the case below, red).

But there’s a problem because our slices have only been colored for the current value of Color (i.e. Red). When the color is changed to Yellow or Green, then it will be set back to default colors. This is where we need to use the NPS Test value. Make sure NPS Value is set to “Test Value”, then change the value of NPS Test so that it changes the value of Color. For instance, a value of 50 will change everything to Yellow.

Set the colors again as we did above.

Do this once more for green.

Now duplicate the 0 pill on the rows shelf to create a new axis. Remove all pills from the marks card on this new axis. Change the mark type to “Shape” then select the shape created for the key. Adjust the size a bit so the shape is smaller than the pie chart on the first axis. Then make it a dual axis and synchronize the axes.

Finally, drag Value Bad, Value Normal, and Value Good to the label card of the second axis. Set the label to be aligned center/middle, then click on the label card to edit it. Add color to each field and place them side-by-side like this:

Only one of these fields will ever have a value—the other two will always be NULL—so only one will ever show a value. (see Dynamically Control Formatting Using Multiple Calculations for more details on this technique).

Finally, change the NPS Value parameter to “Real Value” and filter on one of your customers. My final workbook is below. Feel free to download it and use it as desired.

Ken Flerlage, July 25, 2017.

Hire Us! | Twitter | LinkedIn | GitHub | Tableau Public