Introduction

Text data is an untapped goldmine: by using text categorizations, Al can predict NPS (Net Promoter Score) even better than with quantitative questionnaires. NLP analytics had been used for auto-coding with great precision. Results shaped Sonos' product development strategy.

NPS open text responses are an underleveraged asset. When text categorizations can predict NPS better than quantitative questions, this means that a simpler two item questionnaire can have more validity than a decent quantitative questionnaire. We learned that with help of NLP text analytics software, even manual coding of text data can be automated with expectable accuracy.

Overall this means better validity, faster field times and lower costs that enable full scale surveys. Those full scale surveys are what the inventors of the NPS had in mind - to have a tool that communicates to customers that the company cares, and to have a tool that serves as a constant voice of the customer and tells the organization what the customer cares about. NPS is meant to be the enabler for customer centricity. It only had one challenge...

The challenge of NPS surveys