The big deal with Big Data: Interview with Russell Marsh, Global Chief Data Officer at IPG MediabrandsAdd bookmark
If you have a LinkedIn profile, or pick up a newspaper, or watch the news, the chances are you’ve heard of Big Data. The buzz surrounding this revolutionary new information source has built to gargantuan proportions, however in correlation skepticism has also began to build in some quarters.
Russell Marsh, Global Chief Data Officer at IPG Mediabrands, talks to us ahead of Chief Data Officer 2014 about this building hype, the dangers of underestimating the complexity of the big data puzzle and predicts the path our new digital fountain of knowledge will take before we’re able to truly tap its potential.
PEX Network: We’ve had lots of data within organizations for decades, probably more than we’ve ever been able to process. What do you think has changed now; why are companies really starting to invest in this concept of big data?
Russell Marsh: I think it’s pretty simple. If you look at what Moore's Law has done to computer technology, we’ve reached a point where the cost of processing power - combined with cloud architecture and infrastructure and the development of things like Hadoop and big data platforms - means that it’s no longer cost prohibitive to store vast amounts of data.
Now people are starting to look at what they can do with all this data. The processing power is there to allow us to crunch vast amounts of information and the cost of entry, given that it's open source, is pretty much free. That's why I think there’s an interest in doing it now and why it’s happening at this moment and I believe it will only accelerate.
PEX Network: What are the biggest ways that you see big data changing companies today?
Russell Marsh: First of all, it's forcing companies to pay more attention to their data. It’s also forcing companies to look at themselves structurally because just collecting data is straightforward but if you’re not then using that data it becomes an overhead. So you’ve got to look at your organization and realize that data is, first of all, valuable, and secondly, you must understand which parts are valuable. Thirdly, you need to look at how you are going to use that data moving forward.
It’s forcing organizations to re-assess all of these different components, all the way down to production lines, to understand how they can efficiently optimize and use that data to benefit the business.
So it’s having massive impacts. It’s starting to change organizational structures, not only just so that people are just using it from a data analytics standpoint; it’s changing how they build their businesses and how they restructure ground-up.
"Just having the information is expensive noise. You have to really look at the data that you’re collecting and the data that you have, and understand what’s the value, what are the pieces, the components within the data that I have that when I put them together add value into my business."
PEX Network: That’s a really interesting point; it’s fundamentally changing the way we operate. But while a lot of companies are jumping on the big data bandwagon, there are probably still a lot of people, whether boardroom executives or frontline workers, who remain skeptical about the value. Is that something you see, do you think people out there are still a bit skeptical about the value of big data?
Russell Marsh: Absolutely and I think rightly so. I think big data is such an overused term that it means nothing. Data’s always been big. Even when I was a kid and playing with computers where I used to fill up my floppy disks and I had drawers upon drawers of them. Now you have large organizations collecting terabytes of information, just having the information is expensive noise. You have to really look at the data that you’re collecting and the data that you have, and understand what’s the value, what are the pieces, the components within the data that I have that when I put them together add value into my business.
I think people forget that. I think a lot of organizations at the moment are getting caught up in ‘we’ve got the data, we’ve got to have it’. It’s got some value but they haven’t worked out where. If they don’t find the value, they end up just with what are essentially archives of floppy disks stored in drawers...it just happens to be very big server farms that they’re saving them on. They have to spend the time and take the time to really understand how they’re going to use it, and where they’re going to use it, and they bought the right in most of these organizations to question how and where that data is ultimately going to add value to the business. If it’s not being used in any sort of valuable way that you just shouldn’t collect it in the first place.
PEX Network: How do you overcome that skepticism and can you give an example from your own experience?
Russell Marsh: The way I’ve done it in the past is by starting off very small and taking examples of specific data, data steps or data points, showing how they can be combined, and then how it results in either faster results or it results in new insights that no-one’s ever considered before.
An example of that is at a client where we were handling up to 12 billion rows of data and we were doing data analytics to try and come out with some predictive algorithms out of the back of that. It was taking us about 21 days to run that calculation on a single server. When you factor in how much a license costs and how much someone’s time costs, that’s pretty considerable.
When we then looked to do that on a Hadoop-based platform, we built just over a thousand servers running in the cloud to run the same algorithm and that took one hour. The cost, because you’re only paying for the cycles of the processor, it only cost us something like £500. So the cost differential compared to the power that you’re able to get to crunch the data, and the speed difference in terms of what you can actually do with it is enormous.
Those types of experiments generate proof points that can have massive impact in how you can change the perception of both skeptics and the board, and the business as a whole.
"We’re going to go through a period where these platforms, people realize that it’s a little more complex than they first thought in terms of what you could do, but again without a shadow of a doubt it will come out at the other end and start to add value."
PEX Network: There's risk with new technologies that because of the hype there are unrealistic expectations that the technology can’t possibly live up to. How can companies avoid falling into what a lot of the analysts call a "trough of disillusionment"?
Russell Marsh: So the "trough of disillusionment" is part of the hype cycle. It’s a very well-known and established model where as new technologies come out they hit the press, people get very excited about them and talk them up. They go up this very steep curve where it’s the hot topic of the moment and then – it happens to everything – it comes down the other side where people realize that it can’t quite do what you thought or what people were talking about that you could do with it, and it drops into the trough. Then as people, over time, start to really properly understand the technology, it starts to add value.
It’s going to happen with big data. We’re going to go through a period where these platforms, people realize that it’s a little more complex than they first thought in terms of what you could do, but again without a shadow of a doubt it will come out at the other end and start to add value.
You have to ignore the hype and you have to focus on what you’re trying to ultimately deliver. That comes back to the value piece that I talked about earlier: what is the single points of value that are being added to the business? As long as you stay focused on that and you’re true to how you’re using the technology, you’ll get support all of the way because you’re not building up or over-hyping or raising internal expectations about what you’re trying to do.
PEX Network: My final question is a much broader one. The role of the Chief Data Officer is a relatively new one - how would you define it?
Russell Marsh: For me, the role of a Chief Data Officer is really about connecting the organization and collecting data together in order to drive insight and value into a business. The key part of the role is that connectiion piece: connecting the organization from the marketing groups all the way through to the C-level executives, down to people on the floor so that they understand how the data ultimately will add value to what they’re trying to do and how it can ultimately direct speed into their business and make their jobs either easier or it will add more value to what they’re trying to do for their clients.
It’s more than just a mathematical technology geeky part of the business. It’s really a people business. You’re trying to help people see the value of the data because most people don’t understand it. You have to constantly take very complex ideas and concepts and simplify them so that all parts of the organization understand.
Once they’re connected, connecting the actual technology and the data pieces is pretty straightforward. But if people don’t understand it, they’re never going to use it and they’re never going to buy into it, and I think that’s a key part of what the chief data officer role is at at the moment without a doubt.
PEX Network: That’s a great point, it sounds like the role is as a translator.
Russell Marsh: Yes, a translator. You’re part marketing department because you’ve got to teach people what it is. You talk about Hadoop, people have no idea, you talk about big data they sort of know but don’t really understand. You’re talking about databases that people don’t understand.
You have to translate it, simplify it, put it in context, tell it in stories, show people, build proof of concept so that they grasp the possibilities. At the same time as doing the simplification, you’re also having to deal with the real complexities of how technically you start to do this. It’s a very challenging but very, very exciting and interesting role at the moment.