 Hi everybody, we're back. This is Dave Vellante with Jeff Kelly. We're with Wikibon and this is theCUBE, SiliconANGLE's production. We're here live at the Tableau Customer Conference. theCUBE is a live mobile studio. We go to events. We extract the signal from the noise. We've been using that term long before. Nate Silver, by the way. And it's been our running joke all week. We did not rip him off on that. We started 2010 with that, I believe, Mark. And so we bring you the best guests that are at these events. George Matthews here. He's the president and COO of Altrix, a strategic analytics platform. George is a CUBE alum. Welcome back, good to see you again. Great to see you, Dave. Thanks for having me. So you're telling us off camera. You guys have some serious momentum going. What's new since the last time we talked? Well, a lot of things have kind of evolved in this space, particularly in the last two years. And what Altrix is being able to do is capitalize on how that's evolved. Basically, empowering the user, empowering the data that surrounds the user. So a few of the things I had actually talked about in the CUBE in previous discussions are really coming to shape and form today, where Altrix is now growing and leaps and bounds, largely with partners like Tableau, Teradata, CloudEra, that's really helping us drive that real do-it-yourself movement in data. Yeah, I mean, last time we talked on the CUBE, George, it was, you know, there was a lot of, you know, exhaust around the CUBE, around big data. You know, it was exciting, you know, very exciting times, you know, it went from sort of, you know, what is it, you know, what's the potential of it to, okay, how do I actually do it, how do I get started? So it sounds like you're actually starting to make money at this, getting traction, helping people extract value from data. That's right. Talk a little bit more about that evolution and where we are today. Well, I think I look at it interestingly to think of an eight silver analogy. It's almost the third inning of a ball game, particularly when it comes to data and particularly the use of data inside of organizations. So where we saw data and particularly the lot of the earlier evolution around big data occur in the past two or three years was how to get that infrastructure right, how to be able to get almost petabyte scale applicability around your data delivered inside of an organization. And now there is a real deep question on how do you analyze it? How do you get better context? How do you make better decision around the large sets of information that you have to manipulate, that you have to manage inside of an organization? And what we found to date is that it's not just only the size of that information, but also the variety of data that you can very naturally and seamlessly have to blend together to make better analytic decision. And we really help end users, the data analysts inside of an organization empower them with that ability to blend data at scale. And that's really where Altrix's growth is occurring. So that's really, you're sort of your purpose, right? I mean, that's really why you guys started the company. Now having said that, in this big data world, everybody's big data washing now. Everybody's on the bandwagon, large companies, small companies, mid-sized companies, just really pushing that mean. So how do you differentiate what makes you guys unique from everybody else that uses similar messaging, right? Sure, so I think one of the things that we're seeing is that a lot of the focus and attention around big data again comes back to that infrastructure and being able to build that infrastructure for scale. And we looked at it differently and we said, you know what, it's really around the usability of how a data analyst actually functions. In fact, we're at the Tableau Customer Conference today and where Tableau has seen a lot of their tremendous growth in the market is because they really did focus themselves on the usability of a dashboard, the usability of being able to visual analysis. And that's the same exact thing that Altrix stands for. We stand for the ability to do data integration, data quality, predictive statistical modeling without having to be a programmer, without having to be someone that relies on that infrastructure. Instead, they have all the power to do with themselves with the right tools in their hands. You know, the notion of data quality is an interesting one. We were at the MIT Information Quality Symposium back in July and there was a big theme around the chief data officer and a data czar, if you will. Have you seen that kind of thinking take hold within organizations or is that CIO's job or is it sort of an ad hoc position where a champion will step up? You know, an analytics person will step up. What are you seeing in the customer base? Yeah, so when I look at my customers today, it's interesting that the data role is actually much, much closer to the line of business than it is. Inside of what would be a traditional CIO or IT function. And so when customers like Walmart, for instance, today use Altrix and Tableau together, what's really fascinating is it's the global consumer analytics team. And that's a line of business function that really operates on the extent of all of Walmart's data to figure out what the next best operational use case is for a store opening. Being able to get syndication around their merchandising optimization, their category and their SKU management. And largely those are business users, right, that are making better merchandising and marketing decisions using their data, having tools like Altrix and Tableau at their fingertips. So I really do see this as a movement that's occurring by the line of business user, the person who has responsibility for either driving or making the next best decision inside of an organization, more so than I see it as a technology and IT function. I mean, I would agree with that. And I think the CIO's got a lot on his or her plate. They're generally incented to keep the lights on, really, and make sure the system doesn't go down, make sure email doesn't fail, make sure there's quality of service that's being delivered, that the applications are aligned to the business, that it feels like if you're going to be a data-driven organization, you've got to have a data-oriented strategy, maybe separate from the information technology strategy, but very few companies have that today, would you agree? A data-driven strategy has to be very close to the user, and the lines of business that are actually driving revenue inside of an organization. From the data, right? Absolutely, right. So what kind of people, so when you're going into a new customer account, who are you typically selling to then? It's the business side, it sounds like. Are you seeing, how are you seeing your customers in terms of how they organize their analytic talent? Are you seeing a lot in the business units? Or more centralized, how are you seeing that? Yeah, I know, it's very decentralized today. It's very much to the line of business user. If you see our typical profile of people that are using Altrix for instance, it's a data analyst, it's a business analyst, it is someone who is responsible for data management, data science, inside of a line of business whether they be responsible for marketing activities, merchandising functions, real estate, RF engineering, again, these are all functions inside of a line of business whether that be in retail, telecommunications, healthcare where we actually serve those data analysts at scale. So it is a much more decentralized world when it comes to data analytics today than I've frankly ever seen in like, at least a decade or so, Jeff. So are these data analysts that are using Altrix, they're inside the business units and are they typically building applications that are then rolling out to other folks in their department or in their area? So kind of a data guru within business units who then kind of tries to empower the rest of the organization or their department I should say? Yeah, oftentimes when you look at some of the retailers that we serve, it's a centralized merchandising analytics leader or it's a centralized marketing analytics leader that's creating the first set of insights and then once they have repeatability and scale and how that function works, they're actually pushing that out to say for instance, a store operations lead that might be local to a store in Louisiana. And so that kind of experience of having a centralized thought process that is occurring from a line of business leader and then being pushed out to describe and improve operational efficiencies and as well as just analytic and decision-making efficiencies inside an organization is very much a distributed concept inside of most line of business organizations today. So we know speaking to Tableau executives, one of their strategies is to kind of that land and expand strategy that give them to an organization, maybe one or two users in a department, spread through the department, maybe other departments might see what they're doing and say, hmm, I have an application for that technology. Do you find a similar type of pattern emerge with your customers? Yeah, yeah, when you look at our customers today, it's exactly that same dispersion pattern where initially you'll have a few licenses of altrix largely in the hands, subscription licenses of altrix largely in the hands of data analysts that have a specific problem where they can't blend this data at scale very fast. They want to be able to output that insight into a predictive model, be able to then push that as a package application and they don't want any other individual or a set of groups involved with their effort, they want to be able to again do it themselves. And so that notion of enabling to do it yourself analysts, which I believe now there's somewhere in the neighborhood of two to three million data analysts that really need better tools in their hands, particularly in light of how much big data has kind of emerged as the background they have to operate in is really where altrix finds ourselves growing and helping our customer base. Well, it's an interesting contrast to some of the more traditional approaches where a company will go in and buy a set of licenses for a more traditional business intelligence tool and rather than it spreading, they actually languish, people won't actually use the technology. It's kind of the complete opposite here that we're seeing, which I think says a lot about the usefulness of tools like to have low altrix and how they're really touching a nerve with end users who've been waiting for something like this, but have been frustrated and focusing their efforts on Excel and looking for a better way to do this job. That's right. And one of the things that we see in that experience for those data analysts is not just the fact that they need better tools to do their job, but the amount of information that they have to now take advantage of that's outside the four walls of an organization is pretty substantial. So when you look at the amount of social media data that exists in Facebook or Twitter, the amount of information that comes from syndicated sources from a demographic standpoint from a pharmacographic perspective. Interestingly, one of the articles that was most recently written about the 2012 elections was how the Obama campaign itself was very, very smart in terms of how it used external data. And so one of our customers slash partners is a company called Rentrack. And Rentrack effectively consolidates multiple trillions of data points, effectively on the viewership habits of the TV watching audience on a set top box level at the 127 million households in the US. And so we're talking about hundreds of millions of set top boxes deployed inside of households and their viewer preferences, their behavioral patterns are actually matched to demographic insights. Well, it's interesting when you're actually able to combine those two sources together, you can actually create a much stronger viewpoint in terms of where you do targeted micro marketing as well as advertising. Interestingly, the Obama campaign took key advantage of that data to place TV ads in the right time, right spot at the right maximization of viewership in comparison to doing a broad land of national TV ads that were done by the Romney campaign. And so that actually made a traumatic difference, particularly in the last month and a half of the elections. And so turns out Rentrack is one of Altrix's biggest customers because they actually have the issue of being able to scale and process data from set top boxes and combining that with that external data being demographically oriented. And you guys have this, the analytics gallery in your site, this sort of endium of different apps and the set top box analytics is one of them. That's right. And you're right. I mean, Nielsen ratings are taking a very small sample tiny, less than 1%, one 10th of 1%. And not capturing whether it's YouTube or replays on HBO or, you know, my son watches The Walking Dead, they don't pick that up. And so it's a whole new set of metrics that doesn't rely on sampling anymore. It relies on sort of entire data sets. That's right. I mean, at the end of the day, certain business models, particularly that have relied on sampling as an approach, whether it be survey research, whether it be a sample set that goes from a Nielsen standpoint for understanding viewer behavior, those are all dead, Dave. At the end of the day, like, if you can have the ability to analyze the full extent of viewer behavior across the entire 127 million households in the U.S. and what they're watching exactly at what time and land across not only traditional national TV, but also cable networks and viewership that's occurring over web channels. You know, that's the model that's actually going to create a better experience for not only the user, but also targeting them from an advertising standpoint. Yeah, I actually used that same phrase up at the Vertica conference. I said sampling in a lot of use cases is just dead. It's up on stage with Kurt Monash. He bristled at that, I don't know if you know Kurt and we had a little spat. Yeah, and I think there are cases where sampling makes perfect sense, right? Because there might be a fair degree of situations where you can take a sample and understand that it can be applied to the full populace. But when you have such a dramatic differential and what a long tail basically looks like in terms of TV viewer behavior, then sampling makes zero sense, right? Because you actually want to capture the differential data points along that continuum. Obby Metta made that same statement back at the Hadoop world in 2009 or 10. So basically, he's saying sampling is dead in my business at the time he was with Bank of America. They were doing fraud detection, I believe. That's right. What's the point? Taking a few samples and then six months later telling a customer, hey, you might have been, you know, hacked versus swiping a credit card on your way out of the store and oh, wait a minute, could be a problem. Yeah, anomaly detection when you look at fraud management, in those cases, yeah, I mean, why in God's green earth would you ever sample it, right? You'd actually want to find the pinpoint situation when something like that occurs as opposed to having a small percentage and potentially catch when that instance of fraud occurs. I asked Nate Silver about this notion of crowd spotting and using whether it's social data or, you know, data from the crowd and to do predictive analytics. And certainly we talked a little bit about prediction markets. You know, I mentioned, I made the statement that the public handicappers are actually very efficient, you know, at picking winners, even though a horse race a favorite will only come in a third of the time, but a two to one will come in more often than a three to one and a four to one than a five to one. They're actually quite efficient at handicapping, but he said that he was not optimistic about using social data to predict in the crowd spotting, to predict patterns and do predictive analytics. I'm not sure I totally agree with that. I wonder if I get your thoughts on, if you're seeing customers use that type of information, crowd spotting, social data and the like to do predictive analytics. Yeah, right now, not so much from a predictive analytics standpoint, you can do a lot of great sentiment analysis today, particularly around social media data, but I think one of the bigger challenges with social media data today is that, particularly when you want to create a view of what's happening when and where, a lot of that locational information is actually not being shared by the user, right? I think if you look at Twitter, if you look at Facebook today, about 1% of the tweet stream, 1% of the Facebook feeds are actually inclusive of locational information, right? Because people aren't putting that information alongside of what they're actually doing. And so if you had a richer palette, and I think that's kind of emerging, where as more people are on mobile devices, more people are actually driving their social media through not the traditional sort of lockdown, sit-down PC, but in a more device-oriented way, then the ability to do predictive modeling becomes more interesting, because you have a much richer tapestry to work off of. So I think it's a situation which you'll see a better evolution of the usage of social media in the next few years. But right now, I would agree that it's just really focused today on sentiment. And sentiment, you can get some pretty interesting insights from a sentiment standpoint. One of the things that you could actually do is you can go to the Twitter app on the Altrix gallery and literally run the Twitter application to understand sentiment for any keyword. So if you were actually running that application and, for instance, typed in Apple, you can actually see the entire long tail of insights that people are talking about today of all days, because September 10th is where they're launching their new iPhones into the market. And you can get a pretty interesting indicator of sentiment. Can you necessarily predict what the behavior is going to look like in the not-to-distant future? Probably not enough data yet to do that. You could probably pretty accurately predict when Apple is going to make an announcement. That's been sort of demonstrated. And how are you doing that? You taking the whole fire hose in? We actually are in the public application here taking just the public API of the first 200 feeds in. But actually, our customers are doing the same application for Twitter analysis where they're connecting against the entire fire hose. And one of our great partners, Gnip, as well as one of our future partners, which is here, Datasift, is actually providing the fire hose. And both of them are now packaging, interestingly, a sample of the fire hose, a 1% sample of the fire hose, to be able to accomplish that. So I'm going full circle on the conversation. But that's cool. I want to come back to that because John Furrier's not here. I'm sort of sitting in for him as the anchor. But one of the things he did to pick up on your point is that the metadata, it exists, but it's not accessible, even in the fire hose, right? So what he did to infer location is he developed an analysis of hashtags around NFL teams and college NCAA teams. He's analyzing that now as an indicator of location and also allegiance, trying to find intersection between tech executives, CIOs, for example, and 49er fans. That's right. One of the things he found was that Patriots fans have more CIOs in the base than 49er fans, which was sort of an interesting finding. But so there are, we think we're learning how to infer from that data. Actually, that's a great point, right? Because right now, even though the inherent check-in or the inherent tweet itself might have the geo-located information in it, the information that's inside the tweet or, more interestingly, on the profile of the user, you can actually congregate that together very effectively to derive what a location is around what the insight is that they're driving from a social media context. My first opinion, George, I take on this, is the data is there. We just haven't figured out how to use it yet. It's almost like the old story when TV first came out and the radio guy said, why would he want to watch a bunch of guys doing radio on television? It just doesn't make any sense. And they didn't understand that the medium was going to drive changes in behavior. A lot of people today are looking for an email blast analog for social. How do I connect with these people? Well, how do I, can I broadcast to them? And spamming doesn't work on social. So I think we're, like you say, it's the third inning. It's probably the national anthem on really how to exploit a lot of this stuff. But so I think there's a lot of potential. Yeah, I think so, and I think that a lot of the way solutions are going to be found is actually not too different from how Christian described his keynote yesterday, right? When you look at the notion of being able to find, when you look at the notion of being able to shift, people are actually going to be iterating on this for quite a few years right now, right? And to the point of the third inning of the ball game, right? We're going to find some pretty compelling discoveries, particularly as social media, information that comes from syndicated content, information that comes from the four walls of an organization just describes better customer analytics. And it's happening as we speak, right? There are amazing companies across the board that are basically driving new insights around customer analytics today, using platforms like Altrix, using a visual environment like Tableau to very naturally have the next new insight being derived without having to necessarily be stuck to the rigid formats that they had in their previous existence. Excellent. All right, George, well, listen, always a pleasure to talk to you. It's always fun, yeah. Good to see you again. Thank you again, Dave. Appreciate it, thank you. And congratulations on Altrix and the success that you guys have been having. What's next for you guys? What should we be watching? Yeah, so we have some things that we're going to be announcing, particularly at the Teradata Partner Conference coming up here in Dallas in about six weeks or so. So a lot of focus and attention is going on making sure that the analytics scale very nicely. So we have work that's underway that will be announced as part of this next release of Altrix on the scalability side of analytics. We have initiatives in place to be able to establish that experience for the user being deployed at scale to a grid computing model. So having the same gallery experience that you see here be packaged inside the four walls of an organization inside of a clustered grid computing environment. All those things are very naturally now occurring because what we're seeing is the data analysts are helping drive more usage of data and decision making and therefore they want to be able to share their insights to a broader set of users and Altrix is now a very natural place to not only handle the initial analytic design but also the scale and consumption of their runtime. Yeah, do check out that public gallery, the apps gallery on Altrix. It's a very cool little website innovation that you guys have. I love it, so by all means check that out. You're always a fan of it, so I appreciate it very much. George Matthew, thanks for coming on theCUBE. Kristen Shabo is up next. He'll be here at one o'clock East Coast time so stay tuned for that. We're going to unpack his keynote which was fabulous. If you haven't seen him speak, he's just got an infectious personality that is a wonderful individual. I can't wait to interview him. Keep it right there, this is theCUBE. I'll be back. This is Dave Vellante with Jeff Kelly. We'll be back right after this.