 Everybody, we're back. This is Dave Vellante and this is SiliconANGLE.TV's continuous coverage of EMC World. We've given you wall-to-wall coverage of the event. This is our third year here with theCUBE, where we bring in the smartest people that we can find. We extract the signal from the noise and we share with you our audience in real time what's happening at the event and what the leading thinkers are doing with things like data. So this is the data science segment, the big data segment, and we're here with Katrin Rebont, who's the executive vice president of data platforms at Havas Digital Global, a very interesting firm that is based in the U.S. out of New York and in Paris. And Katrin, welcome to theCUBE. Thank you very much for coming on. We're here also with Jeff Kelly, who is the leading big data analyst in the industry. So Jeff, thanks for coming back. My co-host while our colleague John Furrier is out in the West Coast doing the HBase conference. So Katrin, tell us about Havas Digital Global and what your role is there. What's the EVP of data platforms? What that's all about, and then we'll really get into it. So Havas Digital is a media agency. What we do is essentially consult for our clients on media programs across display, search, mobile, and video. And what I specifically do is manage a technology unit that takes care of analytics that are related to essentially all those digital programs. So what we do is we integrate data that is relative to advertisers having ads on search, display, video, social, et cetera. And we look at how these ads perform, how customers react, and we consult with our clients on how they can optimize. Okay, so when I'm online and I'm searching, let's say I want to book a trip, I want to go skiing in Colorado or something, I'm looking around. And then magically that afternoon I'll see a really great offer. And I'll say, well, that's interesting. So that's not by chance anymore, is it? Let's say essentially you click on that offer and you go onto the advertiser's website. You will want to buy that particular trip and you might like or not like what the advertisers present to you on their website. And so according to that, you will either go through with your purchase or not. And what the advertiser wants obviously, what the marketer wants is for you to go through with your purchase and present you with the best offer so that you actually go through with what you want to do so that you get what they want, what you want. So that's what we help them with. So you help them understand if they didn't purchase maybe why or where they stopped and then maybe make suggestions as to what they can do to- How to change their messaging, how to change their website and how to make your overall experience better so that you can actually get what you wanted to purchase. And you provide this as a service, almost like as a sort of ongoing service. Is it a consulting service? How do you engage? It's typically an ongoing service. Most of our marketers engage us throughout the year. So we tend to optimize on going in an ongoing fashion. So what is the platform that you use to do this? Talk about that a little bit. Well, we are based on GreenFlow. We've been in production with GreenFlow for about two or three years now. And that's the central platform that is essentially the receptacle for our data. And we have obviously a web portal that is providing a layer of analytics and visualization that our clients can log into and can monitor and look at their advertisement. Now, how long has the firm been around? The firm as such has been around for almost 100 years. Yeah, so this is, you're using new technology, but presumably you've brought a lot of expertise. You got 100 years of knowledge built up. What were you doing before the sort of massively parallel, green plum-like system to just have a central God box? I mean, what? The particular system we're talking about here, which is Animes, is a 10 years old system. And it was based on another type of technology before because obviously the MPP technology didn't exist 10 years ago. And certainly not to the level that it is developed today. I would say around 2007, with the boom of social networking, we understood that the scale of data that was available today at that point to marketers was essentially so big that the technology we had at that point would not be able to cope with the volume and the complexity of the analytics that we had to do two, three years down the line. So that's the point where we decided to shift. Yeah, okay, so you architected the system such that you could bring technology in and take it out and bring new technology in? Yes, we really architected it in a modular way because we understood that the pace of change was only going to increase. And so that's one of the reasons why we made the choices we made is because we took technologies that play well with another. Because we know that we're going to have to evolve. We're now looking into Hadoop and into harnessing unstructured data in order to be able to understand better what is happening on the social web and what customers are talking about when they are on forums, et cetera. Yeah, so Jeff, we've talked to a number of Green Plum customers and they frequently tell us, and I wonder, Catherine, if this is your experience, that when it comes to loading times, if that's your gate, that it's a very effective platform for doing that, is that how you use the platform or is it a different value proposition for you? So definitely, loading times are great and we've optimized that very heavily. I think what we find the best value proposition is the analytics side. We have quite complex IP in our analytics modules that are built with libraries like R or Madlib that play very well with Green Plum. And so that's really where we feel that the kernel of the value proposition is. That and the fact that it's natively sort of architected to be able to play well with other systems. So tell us about your internal staff and you mentioned some pretty complex algorithms and analytics. So we're talking a lot about data scientists. Who's doing that job for you there? Do you have a team of data scientists? How did that happen, how did that evolve? We have an internal team of data scientists that we have built up over the past three or four years. And we partner with Green Plum and other data scientists outside of Green Plum for very specific use cases because in our experience it is as important to have the technical knowledge as it is to have the domain knowledge. So we're really looking at people who have that exact crossing to address the use cases that we want to address. So we have essentially a network of depending on the times between six and 10 data scientist partners that we leverage when we want to build a new module. Can you maybe expand a little bit on that domain knowledge? So what do you mean by that? What are the mix of skills for your particular use case that you're looking for in a data scientist? Well, it is really in our opinion important that the data scientist has a context of what is the business question that we are trying to answer. In order to choose the methodologies in a way that is appropriated that will actually bring that solution in the most effective fashion. And for that, we look at people who either have a background in digital advertising, usually, or in finance. We find that those are essentially the two types of background that play the best with what we're trying to do. Okay, so you mentioned between six and 10 data scientists at a given time. How have you, you mentioned kind of the background to look at, but have you actually gone about finding these people, I mean, they're in high demand? They're in extremely high demand. Give some advice to our audience out there that might be in that similar situation and just don't really know where to find this talent. Yeah, well, so I think definitely... Or don't, because you want to keep them all to yourself. Or that, you could throw them off the trail. Yeah, that would be a good idea. No, I think so, one of the ways we've gone about it at the beginning is we've partnered with Green Plum and the data scientist analytics labs. That's how we found the first data scientist that really had a real past in advertising. We also, so I obviously spend quite a lot of time in events like this, networking. And that's where I find, you find the people who really have sort of the best fit with what you're trying to do. Literally, it's random conversations. Wow. Katja, what's the, can we talk about the user experience, your customer? So, you have all this data. You've got some serious expertise in mining that data. How much is the customer involved and is there a self-service component of your offering? If I want to change something, how does the customer interact with your team? So, the customer interacts with our team in a continuous fashion, because in the domain that we are working in, needs are evolving constantly. It is not possible to say today what question you will want to answer in even two weeks. Because of the speed of the evolution of your needs. So, we have methodologies in place. We're fully agile and we have customers essentially having touch points with our team every week and iteratively looking at what we are developing for them. Once it's developed, they have an interface and obviously they have a layer of control over the analytics. They can change the way they want to look at the data. They can change some of the calculations that are made on the data and be able to make that appropriate to their internal use. Okay, so can you talk about some of the results that you've seen, I mean, even in general terms, because ultimately at the end of the day, you're evaluated on how effective you are for your clients. You've been around for a hundred years. You must be doing something right. Hopefully. How has this sort of new wave of big data affected your ability to help your clients? Well, I can talk about one award that we won in 2010 for Best Data Strategy that was using our attribution algorithms on one of the most competitive market for search and that helped the travel client understand whether they should or not and how they should use high volume search keywords in their campaigns. They are very expensive keywords and so it's very important to understand what exact role they play in bringing your customers to finding what they want to find. And so that's one of the things that we did. We update our wife by 300% on their campaign. And so you touch all digital media, right? Obviously, we're seeing a big shift from print to online and I imagine, you know, years ago, you were very much into print. So you navigated that shift. Can you talk about that a little bit? I think the digitalization of media touches really all media. It's print, but it's also TV and it's outdoor and it really has to do with the personalization of media. So today, if you walk by an outdoor post, it is possible that that will be personalized to the area it is, which wasn't necessarily possible 15 years ago. I think print is shifting from being paper only as it was to more of an interactive format like you have on Flipboard that allows you to really personalize the content that you have in the same way, marketers need to be able to personalize the content that they bring in front of customers because that's an expectation today. Today, you, me, all of us, we expect that the content that we are going to consume will be tailored to us and that includes advertising. Yeah, and I think that consumers expect that marketing is increasingly becoming a source of value for them as opposed to buy my stuff, buy my stuff. No, it is about giving me the information that I want at the moment where I want. I want to buy a trip now, give me your best offer because that's when I'm interested and that's where your offer actually has value to me. Yeah, the travel is an interesting example. I think that whole industry in terms of going through this now, several times, I think that whole industry is about right for a big disruption, you know? And you guys are at the heart of that. So very interesting case study. Yeah, you can be attending the Data Scientist Summit? Absolutely. Yeah, okay. Were you here last year? Yes, so I was here last year actually. I spoke with one of the data scientists that we partner with on the attribution framework that we have built. It was a fantastic experience and listening to the other use cases was absolutely wonderful. I think it's a great mind mail that has a fantastic equilibrium between inspiration and concrete use cases. Yeah, so I mean this is what I mean by being a source of value. This is great marketing. It's a source of value. I mean, Green Plum's not there selling you at all. It's a bunch of thought leaders getting together, sharing ideas and I think it's brilliant marketing. It definitely is. Because it's quite useful. So Jeff wasn't there last year but he's going this year. I imagine you'll be able to look out for potential new staff. So if you're watching now and you're going to be at the Data Science Summit, definitely want to know. Yeah, I attended last year. I thought it was very useful. I had mentioned to Jeff. I thought it was a Riley class content. I don't know if he'll go to Strada. I absolutely. They're good events. They did a really good job. So, well anyway, congratulations on the success of your platform and your role. And good luck with finding even more data scientists. Thank you very much. It was a pleasure having you on theCUBE. Thank you for having me. Thank you. All right everybody, we're going to be right back. This is segment three of the spotlight. We're going to go drill down and we'll be right back. We're live from EMC World 2012. Keep it right there.