All financial service organisations rightly place a core focus on putting their customers at the heart of what they do. A fundamental component of this is ensuring complaints are kept to a minimum. According to the FCA, the second half of 2018 saw the number of complaints fall by 5%, from 4.13 million to 3.91 million, which sounds like fabulous news for the industry.

Delving a little deeper into the figures and stats, one notices that all is not as rosy as it appears. Indeed, whilst the number of complaints has decreased from 2018 H1 to H2, from 2017 H2 to 2018 H2 it shows an actual net increase from 3.76 million to 3.91 million.

So what are the main drivers of this? Let us go back to the current figures from 2018 H2. Excluding PPI, the top complaints are about current accounts (15%), credit card (10%) and motor and transport insurance (7%). Moreover, the number of customers irked about credit cards and motor and transport insurance has increased by 10% and 13%, respectively, compared to the first half of the year.

What about the cost of complaints? Surely, they had decreased? Excluding the compensation paid for PPI complaints, a total of £267 million was paid in 2018 H2, an increase of £8 million compared to £259 million in 2018 H1. The costs had surged for companies at a time when they face significant cost pressures. These include an uncertain economic environment, a demanding regulatory agenda and increased competition from the fintech ecosystem that seeks to take on the “established players."

What are the impacts of this?

Well, for starters, more complaints mean more unhappy customers, which in turn corresponds to lower customer satisfaction. This situation is likely to be exacerbated by this newfound love affair we all seem to have cultivated for sharing our experiences – good, bad or sometimes plain dull – to the whole world, or at least our all-important followers on social media. Hence one bad experience can now reach many, many more people. And too bad if those reached have similar bad experiences with the same bank.

Why are complaints hard to analyse?

It is evident complaints are bad for business. Little justification is required. So why are organisations not doing more to tackle this? In an era where terms like ‘analytics’ and ‘machine learning’ and ‘turning data into insight’ are the latest buzzwords, why are organisations not using all this goodness to better understand what their customers are grumbling about? My thoughts on this are not restricted to complaints but also the text data organisations now have, be it a complaint, general enquiry, webchat or, indeed, social media comments.

Text is messy

I believe there are two fundamental reasons for this. Firstly, text data is difficult to analyse. There are large volumes of it with very inconsistent formats and lots of misspellings, as well as the use of slang or "text" talk.

If we consider typical structured data, it fits into nice neat columns and can be analysed with ease. Text is made up of words, and there are 171,146 words in the Oxford English Dictionary, of which the average native English speaker has a vocabulary of 20,000-30,000. That’s a lot of words! Granted there are rules for constructing a sentence. And whilst the average customer may just write "I am unhappy with the service," there are 28 different synonyms for the word "unhappy" in Microsoft Word. Unhappy could easily be interchanged with displeased, annoyed, upset, angry, disappointed, sad, etc.

And then there are individual styles of writing, which can vary enormously, including tone and the use of sarcasm, for example. Someone who is equally unhappy may write, "You can imagine how happy I was at the service you provided." And, finally, interpretability. How one interprets an email can be very subjective.

All financial service organisations rightly place a core focus on putting their customers at the heart of what they do. A fundamental component of this is ensuring complaints are kept to a minimum. Click To Tweet

Because text data is not as easy to analyse as numbers, most organisations do a couple of things: a count of the number of complaints for reporting purposes and then manual analysis. Manual review is very time-consuming and inconsistent. And by reading a sample, you can easily miss the big picture.

Implementing new ways of working is never easy

Secondly, there is a barrier that is less about the analysis of unstructured data and more about implementing new tools and ways of working more generally. Often changing the current way of working can be very difficult. To implement new software, organizations need buy-in from the business team, analysts and IT. Aligning those three can be very difficult. In many cases with technology, it comes down to:

• Data: Can you capture and cleanse the correct data, then provide it in a compliant way for business use? Often this is the biggest challenge.
• People: Does the right skill set (more advanced analytical expertise) exist in the organisation to implement new ways of working? In many cases, there are also cultural barriers.
• Process: You need to think through processes across enterprise organisations covering both the IT department(s) and business teams.
• Technology: Is the correct technology deployed, or perhaps more importantly, what is the right technology and how does this fit into an organisation’s broader analytics ecosystem?

Trying to implement a workable solution is not always be easy. In my next blog, I will look at how text analytics works and share some examples of success. For more information on reducing customer complaints and improving customer experience using SAS Visual Text Analytics, see attached.