How to Recode Net Promoter Score (NPS) Variable(s)

The Net Promoter Score is most people's go-to measure for evaluating companies, brands, and business units. However, the standard way of computing the NPS - subtracting the promoters from the detractors - is a bit of a pain. This article describes how to automatically recode the data so that you can compute the NPS by computing the average, and you get exactly the same answer as you do when recoding manually.

For example, the table below shows data for a particular campsite. If we add up the first 7 categories (0 through 6), 18.13% of people in this data set are Detractors. Adding up categories 7 and 8 gives us 28.94% Passives, and the remaining 52.94% are Promoters. So, in this data set, the NPS score is 34.81, which is OK.

This QScript looks for the likelihood to recommend questions in your project and creates new variables that are recoded to display the NPS score. 

Requirements

• A raw data set that contains a Nominal or Ordinal variable (or variable set if using a -Multi structure) with NPS data. NPS data are questions that contain the words likely or likelihood in the question name, or whose value labels contain the string likely, and which are 11-point scales.
• The Value Attributes in Values for the variables must have the complete 10-point or 11-point scale (note, NPS scores strictly only apply to 11-point scales).

Method

1. Select the variable set(s) containing the likelihood to recommend data from the Data Sources tree. If multiple variable sets are selected, a new NPS variable will be created for each.
2. Click + > Numeric Transformations > Net Promoter Score (NPS) Variable(s).
3. A new numeric variable will appear in the Data Sources tree with the recoded values for the selected variable set(s).

Recoding is performed based on the numbers in the category labels. Those categories whose labels contain numbers less than or equal to 6, Detractors, will be recoded with a value of -100, those categories whose labels contain a 7 or 8, Passives, will be recoded to a value of 0, and categories with labels 9 or 10, Promoters, will be recoded to a value of 100. Categories that appear to be standard Don't Know questionnaire options will be set to missing and will not be included in the NPS calculation.

Drag the variable set onto the page to create a summary table showing the average for each variable in the set. This produces a table that shows the Average which is the NPS score.

You can use this variable set to create crosstabs, splitting the score out for different breaks in the data by either:

1. Dragging a variable set to the Columns field in the object inspector . You can also segment the columns further by creating nested columns in a Banner.
2. Creating a pre-segmented variable set from the Data Sources pane by using + > Filter > Filter one variable set by another. This is akin to making a pivot table where you want the NPS to be the calculated metric, and you want it broken out by different categories in the rows and columns. It's especially useful for visualizing NPS across different data cuts over time. For example, you can filter NPS by an Age variable to see NPS grouped by Age (in the rows) and cross-filtered by a Date variable (in the columns). 

Next

How to Recode Variables Using Category Midpoints

How to Recode Numeric Variable(s) from Code/Category Midpoints