This video series shows how data science can be applied to net promoter scoring to help you gain a much deeper understanding of the factors driving your customer recommendation ratings and how you can take action to influence them influence them.
Introducing Predictive Promoter scoring
This video introduces Smart Vision’s Predictive Promoter solution and shows how data science can be applied to net promoter scoring to help you gain a much deeper understanding of the factors driving your customer recommendation ratings and how you can take action to influence them influence them. Jarlath provides a brief history of the concept of net promoter score before introducing the case study whereby Predictive Promoter was applied to data from the guests of a major hotel chain.
Building a net promoter score predictive model
In this video Jarlath shows how to automatically produce a model that predicts customer’s recommendation scores based upon transactional and operational data. We can also see how we can evaluate the model accuracy and browse its contents to uncover what are the key fields in the transactional and operational data that the model has chosen to make its predictions before finally examining the rules that the model has uncovered to estimate the guest’s recommendation class.
Scoring with a net promoter score model
In this video Jarlath shows how easy it is to take a previously built model that has been generated based on known outcomes and apply it to customer data where the outcome is not known. Using the results from part 2 of this series, Jarlath applies the model to the hundreds of guests who have not provided a recommendation rating so that the hotel chain can estimate whether they are Detractors, Passives or Promoters. At the end of the video we see how the solution was able to identify what the hotel management referred to as ‘Persuadables’ – these were high-spending guests whom the model predicted to be in the ‘Passive’ group. This related to a key objective in the case study whereby the Persuadables would be targeted with a special offer in order to convert them to ‘promoters’.
Text mining and conclusions
In this final part of our predictive promoter video series, Jarlath Quinn tackles the project’s most ambitious task: mining the open-ended guest comments to uncover important insights. Here you can discover how we can use text analytics to extract a series of concepts and sentiments from customer comments in order to categorise the guests’ responses. By visualizing the relationships between these key concepts we are able to see what aspects of the guests experiences are most strongly correlated with positive evaluations. Moreover, by categorising the responses we are able to use the text data to build a model that reveals the hierarchy of factors associated with weaker or stronger recommendation scores. In the final example, Jarlath shows how we can combine the original operational and transactional data with the guests’ open-ended comments to show the interaction between transactional factors such as spending or discount and ‘soft’ factors such as hotel cleanliness or noise levels.