BT’s Chief Data Officer on looking at the enterprise through "data-polarized sunglasses"
At first glance, a plumber and a Chief Data Officer don’t appear to have a whole lot in common. But, according to UK telecommunication company BT’s Chief Data Officer Phillip Radley, there’s a lot of similarities between how they both can seen the hidden "plumbing" of a building or organization.
In this PEX Network interview, Radley talk about all that data flowing around in his organization, describes how BT got involved with Big Data back in 2011, and offers ideas on how companies can avoid falling into the trap of sinking vast quantities of cash in a big data program that doesn’t deliver tangible business results.
PEX Network: You write in your LinkedIn profile that you look at BT through a pair of data-polarized sunglasses. How does the world look different when you’re wearing those sunglasses?
Phillip Radley: If you hire a plumber to do some work at your house, the way they look at your house is different to the way that you look at it. He sees all the concealed pipework; he sees the water tank; he sees the water coming into the house.
That’s the way I see BT. I see the sources of data, in terms of our call centers, our websites, our network telemetry, all flowing around the company. I see it residing in databases and being moved around between applications. That’s what I mean in terms of a data-polarized view.
PEX Network: What is the background to BT’s big data program?
Phillip Radley: We spend about half a billion a year on research and innovation. We have a fairly active stream registering patents and we’ve always done that, being an engineering company.
In late 2011, the team that was working on customer experience research started encountering this big data technology coming out of the Valley. We also have a couple of employees based in the Silicon Valley and they picked up on Hadoop technology.
At the time, they were working directly with business people who wanted to solve business problems. So they looked at product experience, product reliability and were having great success. But they were not really sure how they could downstream that into mainstream IT in the data centers and the data warehouses. So in mid-2012, a program to downstream that into our business-as-usual operations.
PEX Network: What have you found to be the biggest challenge that you’ve had along the way?
Phillip Radley: With a lot of these things, the business doesn’t wake up one morning and say, let’s do big data. They wake up in the morning and wonder about how can we launch our new triple play TV proposition, or how can we make our customer experience even greater. The big investment programs tend to be around those things.
So when you have what is, essentially, a bottom-up technology that isn’t on anybody’s radar and you want to do it quickly, then you have to build a vision quickly and explain how your vision lines up with all the other visions that are already in play.
That was a bit tricky: our data center team had been working extensively on an on premise, private Cloud and they initially didn’t see the need for a separate infrastructure around Hadoop. But we worked with those guys and explained that it would be a better infrastructure; more cost-efficient to do certain types of batch processing and data storage. Once those guys were on-board, we already had some business sponsorship, because they were the ones paying for some of the research. So having overcome those challenges, then we really started to accelerate.
PEX Network: What would you say has been the greatest success of the initiative?
Phillip Radley: One of our big successes around this project is the speed at which we’ve managed to execute. Once we’d overcome some of the initial hurdles, we found that teams were very supportive of this technology. We had really good engagement from a number of CIO teams within various business units and we now have eight project teams working on it, who are already delivering it into their lines of business. I think the big success for us is to do it really quickly.
PEX Network: Speaking more generally, where do you see the greatest potential application of big data, within the telecommunications industry as a whole?
Phillip Radley: Anybody that’s had to build a network management system in a tier one telco, knows all about thousands of events a second coming out of Cisco routers and the huge amounts of CVRs coming off switches, which all need to be processed. So we’ve had this high-volume, high-velocity data problem for a long time.
What’s changed is this new generation of technology, that suddenly allows us to exploit that data and mine it for much, much deeper insight. That network management area is absolutely a key area in terms of that. We can also take it just beyond operational analytics and into really deep, data mining. Network management and data mining are two things that have never really gone together, but we’re now starting to get insights about our corporate network - our IP network layers - that we’ve never had before because we used to just discard the data because we couldn’t store or process it.
The other big area is facing in the other direction, towards the customer. We are increasingly pushing customers through self-service type propositions, initially via the web. This gets us into the world of apps through Android and IOS platforms. And that generates vast potential volumes of data in terms of mobility data and touch points on the website, the so-called click stream data. When you harvest and store that, you get potentially much greater insight into what’s causing customers churn.
Whether you’re a mobile operator trying to wrestle with the data challenge or whether you’re a fixed line operator leading customers to triple play or mobile operators: you really need to have deep insight now, to understand what causes your customers to stay with you, or to leave.
That’s a huge dataset there that has been traditionally very expensive to process. Now you can gather that type of insight at economics of scale, at lower cost, using these big data technologies.
PEX Network: There’s a lot of talk out there that big data has been over-hyped and companies risk pouring lots of money into initiatives that garner little return. How do you make sure that you’re focusing on areas that will deliver the greatest ROI?
Phillip Radley: Early on, when we pitched our vision to the CEO of IT, they absolutely got the technology story and the whole thing about lower cost and more scalability and commodity hardware. But they had that same thing: you’re not just getting a bunch of money to go and build what would potentially be a white elephant.
Instead we put in place a governance structure. We have a forum that meets every two weeks to review candidate projects coming onto our platform. We’ve taken a centralized approach; we have a central, shared platform. We don’t build clusters per project. We have a single infrastructure, so that we can operate it as efficiently as possible and get the best out of the precious skills that we have.
And we have a standard format of a few slides that each project has to present to us. They need to put a technical proposal to us so we can make sure that they’re not using any crazy technology that we don’t support on our platform. And we can check out things, like, how do you know where we’re going to get the data from and how much data do you need? That way we can capacity manage our platform.
Then, we have a finance guy who sits on the governance forum and he says, you need to explain to us why we should do this project, in terms of benefit. Is it going to be a cost reduction? Is it going to improve customer experience? He wants to know which budget and which line of business that benefit is going to show up in.
If projects surpass that Approval Board, then they get to engage and they get help on-boarding onto our big data platform.
PEX Network: How do you see big data impacting the business models of operators?
Phillip Radley: BT is a company with over 100 years’ experience in telecommunications, so we’re arguably one of the first telcos in the world. We pride ourselves on our engineering expertise. However, at the end of the day, the facts don’t lie.
If you have a much bigger data set, which you have mined and you really understand, what is your cost to serve the customer? What is the cost of running an application that supports CRM? What would be the impact of bringing an inbound call center, together with an outbound call center, so you can really examine those business decisions, based on really solid facts that you’ve gathered?
Then you’re in a much better position to transform your business models in ways that we can’t yet imagine.
Within the last two years, we’ve gone into TV, so we now have a 24/7 two channel, TV studio that broadcasts premiership football. There’s a huge investment there. We’ve done a £4 billion investment in fiber-to-the-cabinet and we’re just about to go back into the mobile space. We’ve bought a mobile spectrum and announced some products in that space. So these are huge investments and, if you’re making those investments, you want to be really sure that it’s based on solid data.
Phillip Radley will be speaking at the upcoming Big Data Monetisation Telecommunications conference taking place this November in Denmark. Find out how operators can capture the value and wealth from data and get insight from some of Europe's leading brands.