Measuring your Net Promoter Score (NPS)

[AlteryxAdvocacy]

Author:

Hoss Carroll, Mahshid Borgheian and Bruce Strainer

Role:

Data Scientist, Manager - Customer Insights and Director of Data Science

Company:

Alteryx

Overview of Use Case

We send out an annual survey to customers. This survey has several questions that allow us to derive insights form our customers. A major point of interest from this survey is a net promoter score (NPS) question, which measures one’s likeliness of recommending Alteryx to a friend or colleague. NPS is a metric for assessing customer loyalty, customers that give a 6 or below are Detractors, a score of 7 or 8 are Passives, and a 9 or 10 are Promoters. Using Alteryx, we can create a reproducible and easy to understand process that helps with the entire life cycle of this survey, which includes creating a sample of users and profiling users before and after the survey is sent out.

Describe the business challenge or problem you needed to solve

The results of the survey help drive business decisions, understand our customers better, and determine what we can improve as a company. We use the NPS score as an indicator of business growth and a key measurement of our how our customers perceive us. The NPS score is used as a benchmark for competitors too.

We needed to streamline the entire survey process by creating a reproducible process using Designer. Throughout the entire process we were required to prep and blend the data, create samples, and perform statistical tests. The results of the survey are presented to the Alteryx C-Suite, who in turn, use the insights that are derived in order to drive strategy across all functional departments (product management, sales, community, customer support, etc.). 

Describe your working solution

• The first step in this process is to join all of the different data sources together. This includes Salesforce data and Community data that is stored in our Snowflake database. Once the data is merged into a single source, we removed any Customers/Accounts that we are not permitted to send the survey to.

• The next step was to create a user profile. This profile provided us with the counts and percent of different groups, including regional users, users of specific product, users of specific tools, tenure of users, etc.

• Once a report of customer profiles was created, we needed to sample down the list of users to a more manageable number to support deriving insights from verbatim feedback in the survey. We wanted to create a stratified sample to ensure that certain types of users were included in the sample in order to ensure that we received responses from these users. This process was one of the many ways that Alteryx shined on this project. Using Designer, we were able to quickly and efficiently create the specific stratified sample we needed, which included a somewhat complex criteria. This process would have been extremely difficult and time consuming with alternative data science tools.
• Alteryx made it easy to apply sampling methodology.

• Once we ensured that the stratified sample was complete, Alteryx allowed us to export the list of users in a format that supported integration with Salesforce and send the survey using Survey Monkey.

• Once the survey was closed, we were able to read in the Survey Monkey data and develop reports based on the results. Because of the ease with which Alteryx allows users to create custom macros, we were able to develop tools that provided us with the counts, percent, and NPS score of unique customer groups as well as creating profiles for Promoters, Passives, and Detractors.

• After the survey was administered and collected, Alteryx was used to systemically re-sample the data, providing insights on how the NPS score changed year to year for the different sampling groups and how the overall NPS score varied under different sampling methodologies

• Test of Means tool was used to generate a Welch t-test in order to determine if the NPS scores for different groups was statistically significant.
  

• A custom statistical model was developed leveraging the R tool to help isolate the characteristics of the survey groups from other factors that could impact NPS (such as frequency of use, time using Alteryx or community engagement). The results from the model provided a deeper understanding of the drivers of NPS and actionable insights on how to improve promoters of the brand.
  
  

Describe the benefits you have achieved

• We were able to uncover key insights and drivers about the business. These insights allowed us to determine what actions we can take in order to improve the functionality of our product, our internal sales process, the user experience on our Community, and the user experience with customer support.
• Without Alteryx, a large part of this analysis would not have been possible or would have taken a considerably longer amount of time.
• The consistency of Alteryx workflows allows us to continue applying the same logic and process to future surveys. This allows for consistency across all of the surveys that are sent out, which is extremely important when comparing the survey year over year and accessing the success of actions that were taken as a result of the analysis.
• Moving forward, when we do send out the next survey, the workflows will be easy to understand and repeat.