I have some data that I just used for this from a recent survey. Let me pull it together in a generic format that they can build their viz off of. I would be curious to see how it turns out.
Very kind of you Trudy! Can you let me know how many records you've got? I know they're looking for a large data set so they can demo filters etc. as well.
I also have a lot of NPS data that I could anonymise and share.
Apologies for my tardy reply - notifications are not coming into my Inbox.
Yes, I will get it sent to him today. Sorry for the delay. I just got back from vacation.
Helen & Trudy:
If you guys have some anonymous data set that can be shared I'd be interested in taking a look. We're considering collecting data to calculate an NPS for our company and I'd like to demo what it can look like.
1 of 1 people found this helpful
Here is some sample NPS data and a workbook. I am very new to Tableau, so if any of you have other ideas for how to better visualize this data, that would be great. I got a lot of this from following stuff @SteveWexler posted.
thanks for posting the nps data set. I'll play with it and if I come up with any new ideas will post a .twbx file.
Trudy: I notice your formula to NPS Category ignores scores = 11 and includes scores = 0, but there are no scores = 0. Your data has a 11-point Likert scale from 1 to 11, plus a number of Nulls. For this exercise I'm rewriting the calc to an 11 as a "Promoter," and starting with 1, not 0.
Here's your formula
case [Text Response]
when '0' then 'Detractors'
when '1' then 'Detractors'
when '2' then 'Detractors'
when '3' then 'Detractors'
when '4' then 'Detractors'
when '5' then 'Detractors'
when '6' then 'Detractors'
when '7' then 'Passive'
when '8' then 'Passive'
when '9' then 'Promoter'
when '10' then 'Promoter'
There are two fields that I used for different calculations. One is a value field and the other is the text response field. For this particular question, you will see the values and text response correspond like this:
The NPS Category calculation does include the values for ‘11’, but because I based it on the Text Response variable, it only goes to 10. If you replace the ‘Text Response’ field to instead do it off the Value, then the 11 should be included. This data set also includes those who started the survey, but may not have finished it. On other dashboards, I have this as a filter so we can compare those who completed the whole survey, vs. anyone who just answered that particular question.
That sounds like and looks like you have 2 different fields that measure the same thing in 2 difft' ways. That might be making things more complicated for yourself than needed.
I'll just use the 1 to 11 point scale. I'll post something later tonight or tomorrow. It might make sense for me to do a screen share w/ you to walk you through it. If you want to do that tomrrow, I'm on East Coast time.
I've attached a screen grab of my standard NPS dashboard. It's one in a set of eight dashboards used to present online survey data to clients.
I use the time series viz on the top-level summary dashboard. It provides a "more...." link to this NPS dashboard to allow clients to explore the data.
The NPS dashboard has four parts:
To explain how NPS is derived I show:
- Number of responses (0-10)
- NPS Groups and NPS
In the mid-section, NPS is shown by visitor group.
In the bottom section, I show a time series.
The filters on the right hand side are consistent for the eight dashboards and allow the user to segment the data by a combination of:
- visitor role(s)
- visit reason(s)
- goal achievement
As you can see, each viz is very simple but I hope in combination they tell a good story. I'm interested to see what Steve Wexler will come up with - I use and like his Likert score viz.
I'm constantly tweaking so any ideas for improvement are very welcome . I'm currently working on an interactive XmR chart for NPS (I'm convinced it's the best way to analyse trends).
I took a look at your work book and the .xls file. I attach both with my thoughts on some updates and tweaks.
The biggest change is that I created a single data source in your .xls and ADDED it to the .twbx. Your connection remains for comparison. The new worksheet in your .xls has everything you join together in your original data connection and keeps it to one row for each response, or 2789 rows. Your data connection joins a table where you have up to 4 rows of data for each response. This makes thing more complicated than needed.
There are times we need to reshape the data into multiple rows for one respondent. At the moment, I don't see that need with this data set.
If you want to take a few minutes on the phone so I can walk you through things let me know.
Looks nice. I may try to incorporate some of your approach in the .twbx we're sharing in this thread. However there's no date fields so we can't do any time series analysis.