Case Study: How do I improve NPS and customer engagement?

Challenge:

A leading UK High street Bank has a large quantity of data, however they had no tools derive value or understand qualitative feedback. Heron was approached to help to analyse their data in the aim to understand customer engagement to establish proactive actionable outcomes to improve their NPS score. 

How did Heron approach this?

The data we had was a mixture of online and telephony feedback including NPS and verbatim with attached meta data. We set out with three methods to review the data.

1. Sentiment and Theme Analysis

We carried out an entity and sentiment analysis to show us the key elements within the text unhidden to the human eye.

2. Machine learning pattern matching

Using machine learning based analytics allowed us to look further and develop views inclusive of the meta data allowing us to start to predict patterns.

3. Use of sentiment tool plus post processing collateral

We then took both elements of the data and combined them to demonstrate the power possible by taking both of these processes together.

We took the approach to highlight anomaly and extremity issues, with the theory for doing this that these are the key issues and dealing with them will allow the second level issues to be identified. In order to do this, we began with an overview of sentiment. This set the scene and showed the bias applied throughout to the sentiment variances given through the different feedback approaches. As a response to identifying a potential shift in sentiment we explored a shift to analysis based on happiness instead. This looked at other triggers including anger, calmness and fear to assess a comfort level from customers.

Technologies Used

  • Machine Learning
  • Natural Language Understanding

The Outcome:

Sentiment alone is only a tester of the current position, However through the application of machine learning looking at trend on time, Verbatim review of multiple entities in feedback we were able. with a 97% accuracy rate, predict the movement between positive, neutral and negative sentiment within customer feedback. Additionally, we identified key topics and triggers for positive and negative sentiment which informed the proactive approach to their customer engagement. Following this, our client overhauled their training of call centre staff and re tooled their use of the knowledge base to improve customer engagement