Improving customer experience and reducing costs to serve at the same time sound contradictory objectives. Surely improving customer experience requires a larger investment in it? Yet in some instances, I believe it is possible to achieve both at the same time.

Last year I wrote a couple of blog posts (part 1 and part 2) describing one of the most popular use cases for SAS Text Analytics. In these blogs, I shared my opinion that a narrow emphasis on positive - negative sentiment polarity was not particularly useful. To properly understand customer opinions requires a more holistic approach, which focused on your organisation’s business drivers and objectives, using unstructured / structured customer data and combining machine learning and natural language processing techniques.

I described our approach to Voice of Customer analysis, which helps our customers better analyse their own customers’ feedback and opinions from data sources like customer satisfaction surveys or complaints systems. Ultimately the approach demonstrated that our customers make better decisions to improve the experiences they deliver, if they have consistent benchmarking of experiences and better root cause analysis of issues.

Recent steps with Voice of Customer

I believe that all of this is still valid and in the couple of years since then we’ve continued to work with several customers to deliver exactly this in their own organisations. For example, SAS helps Royal Bank of Scotland unlock the 'Voice of the Customer' hidden in 250,000 web chat conversations per month. RBS focused on improving the service from its webchat function, which resulted in improved agent performance and higher customer satisfaction survey scores. Another expected benefit was to ease demand on the contact centre, because of the more efficient webchat function.

I’m seeing this last point more and more frequently in the projects I work on with customers. In addition to the traditional Voice of Customer benefits of:

• improving customer satisfaction scores, such as NPS
• reducing complaints and customer churn

Customers are now wanting to reduce the cost to serve their customers, as a by-product of better customer service. For example, in one recent project we identified the following customer service interactions in one channel alone over the period we analysed:

• 19% of issues were not resolved first time
• 10% were caused by either a failure to action a customer commitment or to resolve something first-time in another channel
• Approximately 26% of the service requests were for transactions that could potentially be moved to an online digital process.

So, over half of the demand for customer service on this channel could potentially be avoided totally. Using machine learning we then went on to identify some of the key reasons for non-first-time resolution and failures to action customer commitments. Good customer service is expensive to provide, requiring skilled agents. But even if only 10% of this volume could be avoided, this represents a significant opportunity for cost reduction, whilst in parallel improving customer service.

Another customer had an outsourced service centre, just reading and categorising complaints into different categories. The service level agreement required this to be done within three days. This introduced delay into the process of resolving the complaint. Our resolution removed the need for the outsourced process (at a significant cost reduction), improved coding accuracy and meant that complaints were handled quicker with improved root cause analysis.

I will be exploring these ideas in more detail in my webinar Using text analytics to deliver exceptional customer experience on 14 March. I hope that some of you may join me then.