Customer experience executives need to constantly understand how things are changing, in new or existing products and services as well as changes in customer expectations and competition. Current VoC analytics can provide some directional capability such as satisfaction scores and topic trends, but quantifying the detailed topics, let alone finding their predictive factors was not previously possible.
PrediCX changes this. Its unique capability is a live early warning system of VoC insight, classified in a consistent and accurate way. As well as quantifying what is happening in the business and the sentiment of those topics, PrediCX can generate further predictive models to identify the drivers of those insights and their financial impact. It helps operational users quickly prioritise and solve issues, and helps CX and marketing executives identify revenue and cost-saving opportunities as well as take strategic decisions regarding investment and customer journey enhancements.
PrediCX overcomes the many challenges associated with analysing VoC data. Whilst there are plenty of customer experience solutions with sophisticated text analytics, classifying the various complex topics that customers talk about in a consistent and reliable way requires intensive human intervention. This typically means that there are analysts whose job it is to classify data, either by building dictionaries and rules with key terms, or by manually classifying for machine learning. Both techniques are laborious, can be quickly out-of-date, and can miss new issues if new terminology isn’t pre-empted.
PrediCX features two key technologies developed over a decade of academic research. Firstly AIR (Automated Information Retrieval) extracts all of the possible topics and associated sentiment from the raw data. Secondly OL (Optimized Learning) takes this rich output from AIR and uses it to generate predictive models. Crucially, OL ‘asks’ users for specific input for validation where it needs it to optimise performance.
As a result, PrediCX can automatically classify Voice of Customer data to a high degree of accuracy with minimal user input. And because it’s accessed via an API (cloud or on-premise) it can be easily deployed and operationalized. As well as early warning, use cases such as handling complaints, queries and even warranty and insurance claims can also be automated to an optimal degree. It can also be applied to internal stakeholder voice such as technicians and front-line staff.