The problem

The global experience team of a leading North American energy and utility company wanted to listen to their consumer base on social media, to allow them to improve the customer experience and engagement. The voice of the customer – as they call it is their framework to listen to customers on Twitter – and the official website were primarily used for measuring the impact on brand perception and loyalty.

The cognitive solution

The maturity in unstructured data mining is high with the evolution of machine learning, and Cognitive BI, always close to academia, laid down an innovative approach to help out in the agenda of mining rich insights from social media data. Two broad buckets were created on how the problem was approached:

1. Conversational nature

• Collecting and analyzing information on the industry, products and services, conversations on celebrities that drive brand image
• Segregating volume of conversations to topics – products will have a significantly low volume of conversations as opposed to volumes non-client advertising/celebrity directed conversations
• Understanding the generic sentiments, pain points of the customer allowing the team to effectively tackle which ones need attention and to be addressed sooner.

2. Conversational strength

• Building a social graph on the unique participants in the conversations/topic allows to rank a topic and give it a strength based on the depth and spread of the network
• There are varying impacts based on the frequency of posts, time spent on social media as opposed to others. Using this information in the solution architecture is crucial for estimating impact without any bias

Key insights

The deployed system has enabled the team to monitor the customer world in four main themes:

• The privacy and security debate
• Customer support and ease of information mining
• Platform and website pain points
• Payment oriented troubleshooting