Blending Sentiment with VOC: taking an intelligent approach

Back in July, I blogged a potted history of text analytics and its growing role in the business world as the champion of helping large organisations make sense of the vast amounts of text that makes up, depending who you listen to, 80 to 85% of all enterprise data.

With a plethora of potential use cases and sources of value, the stability, reliability & flexibility of the underlying technology is progressing rapidly. The race is very much on for vendors like NICE to maintain our position at the forefront of innovation in successfully utilizing and productizing this hugely exciting technology.


In my role in NICE Fizzback’s Transformation department, I’m lucky enough to advise some leading companies on how they can successfully operationalize the Voice of the Customer throughout their organisations.

And one hugely exciting source of VoC operationalisation that I’ve looked at recently has its roots in text analytics – more specifically in sentiment analysis. This is the ability to assess customer verbatim comments for their underlying tone, assigning each comment a sentiment score. When sentiment is blended with additional VOC data (satisfaction & likelihood to recommend metrics, topics & categories for example) – the operational value is significant – and here is an example:






The bars in this graph represent a Customer Value Index – a metric that looks at how often a customer transacts and the amount a customer spends in the retail arena. We can see in the central three bars that we see the relationship we’d expect – NPS Promoters have a higher propensity to transact more frequently and/or spend more with the organisation in question compared to NPS Detractors and NPS Neutrals.

What showcases an operational use case for sentiment analysis here are the bars to the extreme right and left. When we only consider classic NPS promoters who have given a very positive verbatim comment, we see another layer of complexity – experience thus far suggests a 23% differential in customer behaviour between plain NPS Promoters and “Super Promoters” giving high scores and a highly positive comment. The inverse relationship holds true for “Super Detractors” – opening up the possibility of prioritizing customer recovery efforts towards those customers most at-risk.

This provides just one example of how sentiment analysis can add another layer of depth and value to VOC analysis. The creation & manipulation of this new metadata across millions of pieces of feedback received every year opens up real possibilities for a more intelligent approach to leveraging high-value customers and recovering unhappy ones.

Andrew Robson

For more info on NICE Fizzback, please visit www.nice.com