If you ask someone – a customer say – whether they like your product or service, the most likely answer is: “Sure, it’s fine”. If you ask the same customer to score your product or service on a scale of ‘1 to 5’, the likelihood is that they will score you around a ‘3’.
This is not necessarily because they quite like your product but because it is the path of least resistance. If the customer says, “Actually I hate it and I am about to change provider,” he knows that ahead lies a longer conversation than he wants. He suspects, in his heart of hearts, that you are about to transfer him to that place we all hate – that specialist sales team that will stop at nothing to change your mind. “But, what exactly have we done wrong? What if we offer you this? Or that?”
We have all been there, and if we haven’t we have had nightmares about it.
So the customer says the product is fine, the statisticians say that their customers are basically happy (and aren’t we clever). And the customer churns anyway.
The answer, of course, is not to ask the customer in the first place, but to use science, including – but not exclusively – big data and analytics to find out how happy customers really are. Or, better, use the vast variety of data sources to test whether the customer is telling the truth, or just trying to get you off the phone.
The best approach, according to the paper, is to avoid trying to do everything at once. With CEM projects, and big data projects generally, the watchword is to start small and use common sense. Then expand the scope.
For instance, when the customer says he is happy, you can check how often he has used your product or service in the last month, compared to a few months ago. You can check social media for any comments about your service.
Over and above this, you can check which channels your customer prefers, and use those channels to communicate with him. If he uses the app, send him notifications via the app. The same applies to the web, the phone or social media.
The good news is that this kind of agility is now so much easier with the advent of virtualized network functions and software defined networks.
The message from the paper is clear. Take a deep breath. Then apply common sense, now called business logic (for some reason). And, perhaps most importantly, spend much more time than you think in data preparation – 50% to 70% of the project will be invested in data preparation, says Analysys Mason. Multiple sources and systems will have different ways of displaying data, and is full of – actually quite simple traps – such as different ways of displaying the age of the customer (date of birth in one system, age in another).
The message is simple, and one we have promoted before. Big data and advanced analytics are hugely powerful tools, but only when applied incrementally and with common sense.