The Big Picture for Big Data
Several weeks ago, Information Management had the opportunity to sit down for a wide-ranging discussion with Dr. Rob Walker, vice president of decision management at Pegasystems, the Cambridge, MA-based provider of BPM and CRM solutions. Walker, who holds a PhD in computer science and began his career pioneering predictive analytics at Capgemini in the 1980s, shared his perspectives on the growth of big data and the challenges faced by companies as they struggle to derive real business value from the multitude of data management technologies that big data’ represents.
Information Management: How prepared are large companies to leverage the potential opportunities in CRM, BPM and decision management presented by predictive analytics and big data?
Dr. Walker: The essential value of data hasn’t really changed, but a lot of big data initiatives have been launched for the sake of undertaking a big data initiative — even though there may not always be a very clear business objective.
There’s a lot of confusion about what big data really means and how big it has to be to be big.’ There has always been a large delegation of Fortune 1000 companies that collect a lot of data but don’t do much with it. A lot of companies haven’t given enough thought to making data actionable. They don’t have a clear, top-down objective for what they want to achieve — whether it’s reducing their risk, increasing their ups-ell opportunities, improving customer retention or reducing fraud.
Our perspective at Pegasystems is that you should always follow the business; follow the money as well, rather than just do stuff’ because you have the technology to capture more data.
Some industries have always been a little bit ahead of this curve. Predictive analytics and next best action — which is really about making data actionable and doing something with it — started in the telecommunications industry. Years before the banks started doing this, the telcos were using next best action for marketing. Now you see insurers starting to do this — and it will get into manufacturing; it will get everywhere.
Why did this approach to using data gain traction with the telco industry first? What is it about certain industries that primes them to be early adopters of these technologies while others lag behind?
Originally, predictive analytics at places like insurance companies was not really actionable — BPM and related processes weren’t in place yet. So predictive analytics was used after the fact, primarily for insight to support future decision making.
Telco started gaining an edge when mobile phone service providers were competing feverishly, and suddenly everybody needed these churn models — these attrition models — so they could predict which customers were likely to leave with some degree of certainty.
If a customer came to your website, or contacted your call center, and you already knew that there’s a strong probability of churn — now you had the possibility of making a little better offer than the competition and that was worth an incredible amount of money. That predisposed the telco industry towards analytics — that plus the fact that they already had a lot of data from phone purchases, calling records and so forth.
The industries that have moved first are those that are already collecting the data and have a real pressing need to monetize it.
Banks started moving in this direction as they became more customer-centric in the early 2000’s. Because if what you’re about is improving the individual customer experience, then you have to know what the customer actually wants and that requires a lot of analytics.
Since the 80s, banks have undertaken campaigns to move particular products. From an analytics perspective, all that requires is to determine the customer set that will be the most receptive to that product, which is relatively simple. But for next best action, you suddenly have to calculate the probability of a thousand things all at the same time, because the challenge isn’t to answer who is interested in product X?’, but what should I be talking about with this particular customer right now?’ And to do that well, to make a relevant recommendation, you need thousands of predictive models that all execute in real time at the moment of truth.
Now we’re seeing traction develop with insurers, who have developed more touch points with their customers and so are generating more data about them.
Another way of looking at it is business versus consumer-oriented marketing and sales. Historically, B2B sales were more one off — you didn’t generate a lot of data, so B2C was way ahead in that respect. But now the analytics have improved in the B2B space and a lot more data is collected. We still don’t have the scale that we have in the B2C space, so it’s not as effective, but there are very large companies with a lot of B2B customers that have undertaken these predictive efforts.
For companies and industry segments that are trying to get more traction in these areas—what kind of political and organizational challenges are they facing? What kind of shifts need to take place so that they can respond quickly and flexibly with the actions that are suggested by the data that they’re taking in? To what extent do corporate silos and hierarchies get in the way of this?
Almost without exception, companies are still very siloed. This whole thing about customer experience and customer-centricity really forces companies to become centralized in their customer positioning. This whole concept of next best action really forces the issue — so it’s a very interesting dynamic in many of those companies.
I’ve worked with old line banks in London, where the credit risk department and the marketing department have never spoken with each other, have never been in the same room together for hundreds of years. Yet they share the exact same customers. They both have data; they both have analytics, but they’re completely siloed — so if credit card offers were made to customers by the marketing department, then 65 percent of them would be turned down by the risk people — because the effort wasn’t integrated.
The good news is — because at the end of the tunnel is not just light but a really, really big pot of money — that all sides of these companies are starting to really talk to each other. But for this to happen, you have to move up high enough in the company’s hierarchy so that you’re engaging with people who can really see the bigger picture.