Conversation analytics

Big Data is obviously mind-blowing amounts of digital information.  However, it is also defined by its diverse nature, the speed at which it is created and accessed and the emerging techniques to extract insights at scale.  Together, these characteristics are the 5 V’s of Big Data – volume, velocity, variety, veracity, and value.

Until recently, customer interaction data was ‘Big’ only in terms of the first three of the V’s.  That is, the volume, velocity, and variety of customer interaction data that is created and collected every day through the numerous ways customers interact with organisations.  From traditional phone calls and chatbots to the off-business system complaints and social media commentary.

Organisations have stockpiled vast amounts of increasingly diverse and complex customer interaction data with varying degrees of success. Unstructured data such as audio and text files require different approaches to storage, governance, access, and analysis.  The promised data lakes of unstructured data have often devolved into inaccessible data swamps, resulting in limitations around usage and implementation.

However, in today’s age, technology has finally caught up with the hype.  With advances in secure cloud storage, AI, and machine learning, we have the tools and techniques to interrogate the veracity and realise the value of customer interaction data.

We can now extract and analyse the rich content of customer interactions.  Improvements in call transcription engines have advanced from word salad results to highly accurate conversational text outputs (for a good laugh, google search examples of transcription fails).  We can extract and analyse high quality acoustic features such as tempo, amplitude, pitch and tone to better understand the non-linguistic context of speech and of the speakers' emotional state.

Natural Language Processing codifies interpretation of meaning within customer interactions, revealing intentions and behaviours at scale. This is now used to reduce churn by predicting at-risk customers, provide information about customer satisfaction, and improve operational efficiency through recognising and responding to customer needs.

Customer interaction analytics enhances quality and risk management in an environment of increased public expectation and regulatory scrutiny.  All interactions can be automatically checked against quality frameworks and regulatory requirements. This informs targeted coaching of contact centre employees by analysing their individual performance across key metrics.  Complaints can be identified from interaction content, including those that fall into the expanded regulatory definition that does not require the customer to explicitly state ‘complaint’ or ‘dispute’.  Further, preceding indicators can be identified, which allows for preventative action before a customer complains.  Voice and interaction analytics can detect historical agent misconduct to support regulator directives and remediation programs.  Or, better still, emerging risky agent behaviours such as mis-selling or pressure retention can be detected and treated before becoming endemic.  Customer interaction data now provides direct, real-time access to the feelings, intentions, and experiences of customers and to agents' quality and compliance behaviours.

Consider that for a second. Your business now has direct access to your customers' thoughts and feelings and what your agents are doing in real-time!   So much potential, yet only a select few are truly optimising the untapped value of the customer data they hold.

Organisations must reimagine their customer interaction data assets, recognising they have a direct line to the voice of the customer and start listening to what their customers are telling them.

What is Deloitte doing in this space?

We are working with our clients to help make the most of their Big customer interaction Data, supporting:

• Customer outcomes – examine customer sentiment, behaviour, and emotion across customer journeys
• Risk and compliance management – automate monitoring of agent adherence to regulatory requirements and oversight frameworks
• Agent quality management - benchmark agent performance across automated quality measures
• Complaints management – automate the identification of complaints and early identification for customers who are likely to complain
• Remediation programs – automate the identification of potential customer detriment and vulnerability.

Discover other key insights into conversation analytics and explore the space of advanced voice analytics here.