Tornadoes, or as we call them here, twisters, are a common sight in Texas. 100 years ago, when we wanted to find out if a tornado was coming, we had to either see it, or hear it. “It sounds like a freight train,” my grandfather, who had seen a few of them, used to say. Here in South Texas, we are at the end of the “tornado alley,” a term for the region in the United States that experiences tornadoes most often where last year alone, there were 1,676 reported so called twisters.
We know why tornadoes happen quite well. A large mass of cold air moves over a large mass of warm air, creating unstable atmospheric conditions — warm air, the lighter of the two, moves up through cold air, forcing the cold air to move down and around it. If there is enough energy, these boisterous winds twist to form the infamous funnel cloud.
A few decades ago, tornadoes would sort of “hit,” as my grandfather would say. But today it’s a different story — we can actually see them coming before they happen. And how is that? With data.
Weather forecasting: a success story
Those working in weather forecasting are used to being the butt of a joke. Even Larry David pokes fun at them in a Curb Your Enthusiasm episode, claiming “the weatherman” predicts rain when there won’t be any, so he can have the golf course to himself. But there’s no conspiracy, and despite what most people think, meteorologists are actually quite good at their job.
In fact, weather forecasting is one of the success stories in data prediction. Sure, they get it wrong sometimes — a study of TV meteorologists in Kansas City found that when they said there was a 100 percent chance of rain, one-third of the time it never actually rained — but in the last decades, our understanding of the molecules in the atmosphere, coupled with ever-increasing computing power, has improved our forecasts in every conceivable way, predicting rainfall and temperature, hurricanes and tornadoes with shocking accuracy and detail.
In other areas where we want to predict the future, though, we haven’t done quite so well. Sports analysts, political pundits, investors and economists have a very poor track of correctly predicting the future — think of elections in 2016, when nearly every poll and outlet forecast Hillary Clinton as the new President of the United States, or even a bit further back, when in November 2007, as the bubble was already bursting, experts foresaw less than a 1-in-500 chance of an economic crash as bad as the one everyone would experience just a month later.
We’re terrible at predicting the future
Generally speaking, humans are pretty terrible at predicting the future. For a number of reasons — the biggest one being our own biases. We tend to believe, beyond all reasonable measure, that the things we want to happen will happen. We tend to process new data to confirm our beliefs, many times ignoring the data points that don’t conform to our ideas. We are overly optimistic that bad things only happen to other people, and as a result, assessing risk is not our strongest suit.
And although we can build computers and algorithms that are extremely efficient at synthesizing terabytes and terabytes of data, we’re not. We might notice some trends over a short time, but we’re helpless at understanding long-term changes and statistical trends. As Prakash Loungani, an economist at the International Monetary Fund, told a senior writer for FiveThirtyEight, “very, very few recessions have been predicted nine months or a year in advance.”
The truth is we have gone around for thousands of years using a very loose framework of what’s observable, together with our own intuition. We resort to our experiences and understanding of the world around us to address a current set of circumstances and make decisions on what to do next. In all fairness, our estimates are good enough to keep us alive, but we don’t have enough brain power to compute much else. And that’s where technology helps.
When data is used properly, it can go beyond just informing our decisions. It can actually mask our biases and shortcomings, and show the best path forward. If we can get to a place where we trust data even when it’s counterintuitive, or it goes against our desires to go in a particular direction, we can use technology to help us make better decisions.
But how does all of this relate to our world of customer experience?
Back in 1991, when I first started working in call centers, we were making outbound calls to people using manual phones, but just a year later we got an automated dialer — we would load up lists of many tens of thousands of names and phone numbers, the dialer would make the outbound phone calls, and when it got a live connection, it would deliver it to the agent. It was a game changer, and our team was more efficient than ever before.
But a few months later, as a branch manager, we encountered a problem. We were wasting time calling “bad numbers,” numbers which, for whatever reason, no one ever picked up. So I gathered a few data points and realized that if a number had been called more than 10 times, there was a massive drop in the probability of a person ever answering the phone. And so we developed a model that would then run against each list to remove these numbers, which resulted in a tremendous increase in our efficiency.
That’s when I first experienced the power of using data to improve the way we did our jobs. Over the years, I’ve seen more and more how much it can enhance the customer experience as well.
Improving the customer experience
Businesses have access to an overwhelming amount of customer data — we can literally listen to what our customers are saying to us. And not only that, but we can also understand what they’re trying to achieve, what’s their intent, how they’re feeling about it. And if we play our cards right, this data can offer us some truly unique, enabling insights.
But achieving that goal is very challenging. There’s an explosion of data that is scattered across multiple channels while customers’ expectations are increasing, requiring that companies understand them and engage in more meaningful ways — not to mention there are four very different generations in play in a global environment of complex cultural and socio-economic segments.
In order for businesses to go above and beyond, they need to understand customer past behavior, predict customer wants and needs, and successfully deliver a positive customer experience at each “moment of truth”— key interactions where the customers’ needs are met in a way that builds trust and loyalty.
So how can we replicate the modern successes of meteorology in the customer experience? First, we need perspective.
Meteorologists gather hundreds of data points from satellites, aircrafts, weather stations and weather balloons around the world and in space, to then analyze all the variables in massive super-computing environments. So they have created a unified set of data that enables global perspective, and pinpoint local accuracy. In our industry the parallel would be to build a unified customer view, where all the systems are integrated and all the data is pooled together.
That will give us the ability to see the customer journey as this longitudinal arc —where we can understand which link the customer clicked on the newsletter, see how far they were in their purchase before they abandoned the shopping cart, if they contacted a support agent, or used self-service support instead. We will know where they are going and what they’re saying. And that sets the stage for AI to be introduced.
Once all systems are in place and the data is unified and ready for analysis, an AI-powered model can help us with our predictions, identifying specific business questions or problems and determining the best set of actions to take in order to improve each customer’s experience — whether that’s a change in channel, a proactive message, or a more empathetic agent.
Sentiment analysis uses algorithms to determine the way a customer feels about an interaction, whether it is positive, negative, or neutral. In the world of customer service, this means agents can be provided with behavioral hints to coach them on what to do next or how to react to a customer query based on the sentiment, emotion, and intent of the customer.
Before this, the only way businesses could find out how people were reacting to a particular product or service was to gather focus groups or send out surveys, a time-consuming and frankly inefficient approach. Sentiment analysis can help businesses gain perspective, improving the customer experience and insight into products and processes in near real-time.
In our industry, the reality is that all the data we need is already here. One of my mentors used to say in his raspy all knowing voice, “You’re either reading the newspaper, or you’re writing the story.” While data used to be something we analyzed to see what happened in the past, in our fast advancing CX environments today, we can use data to impact and understand what’s happening in the actual moment of the interaction. We even have the ability to predict future events in the same way meteorologists can predict the path of a twister, or the temperature on a sunny spring day.
The big difference between us and meteorologists though, is that we can change the outcome. That’s good news for we practitioners, and for customers who are looking for that perfect experience, every time.