 Live from Las Vegas. Extracting the signal from the noise. It's the CUBE! Covering IBM Insight 2015. Brought to you by IBM. Now your host, Dave Vellante and Paul Gillan. Welcome back to Insight 2015 everybody. This is the CUBE. The CUBE goes out to the events. We extract the signal from the noise. This is our fourth year at IBM Insight. It's been really interesting to watch the transformation of IBM. IBM had a series of bespoke BI, analytics, information management products. And then several years ago, three or four years ago really, and I'm sure the strategy started before that, decided to put that together in a single focus business unit. I've said, and they essentially super glued that business to the big data meme. Even though IBM doesn't use that big data meme, they call it analytics and cognitive. And they have become a leader now with the explosion of Watson and other analytics. And of course there's a major component of that is cloud. And Joel Colley is here. He's the general manager of Insight Cloud Services at IBM. Joel, welcome to the CUBE. Glad to be here. So tell us about Insight as a service. Everybody wants everything as a service. Our focus is helping people understand the data that's out there in the real world. And in the context of cognitive, you probably hear people talking about systems that can understand the world and then systems that can reason about the world. So understanding and reasoning are the two key pillars of cognitive. My focus is on the first of those. How do you get systems that understand the world? What does it even mean to talk about a system that understands the world? And the key to that is actually the phrase you used a minute ago. It's extracting signals from the noise. I'll get to that in a minute. But it starts with understanding people, places, things and events in lots of different data sources. And there are people, places, things and events in every outside data source, but there's no easy way to put those pieces together. You don't know that this given set of data in one data source that's labeled as Joel Colley is the same as this one around Colley Joel, nor this one around this GM of Insight Services, nor this one about an IBM executive. They're all the same person, but it isn't obvious that that's the case. So people, customers struggle with putting together that outside data in a way that makes it interpolate, understandable. And before you can get any understanding, you have to start with people, places, things and events. So can you add some clarity to what your business unit actually does and sells? Is it software as a service? Is it consulting as a service? The closest analogy would be data as a service. So we bring in data from Twitter. We bring in data from the weather company. We bring in data from about 150 open data sources, including things like bureau labor statistics and census data and lots of other open data sources. We'll be adding more and more data sources over time. And as we bring in that data, we manage the licensing issues, we manage the usage and rights issues, which turns out to be a fairly substantial part of the challenge people face. But the most important thing we do is we sort out these people, places, things and events, and we draw the connections between the different data sources to enable people then, when they want to consume that data, to know that this set of data across these three sources are all related in this way. So it's kind of connecting dots as a service. That's exactly it. Find the dots and make them connect. And you sell that as, what, a monthly subscription? And it runs in the cloud, presumably? It's all cloud-based. There are several different ways of consuming it today. What we announced yesterday actually is that we have API data access to the basic data from Twitter, from the weather company. We've done some things to package those APIs up to make them more consumable for different industry use cases, but it's basically a data access mechanism. Can you talk about this API initiative that's interesting? There's a lot of companies, of course, that would like to get access to Twitter data. You've done a lot with the Weather Channel lately, really promoted that, including at the keynote yesterday morning. How cheap and easy are you going to make access to this data? So it's very easy to get started. It's very easy to do early implementations. The pricing that we end up putting in place for access to the data is very much a function of the industry and the use case that gets applied, and it can range from fairly very inexpensive to much more expensive depending on how much value people are getting out of it. Part of the key of this whole business is actually it starts with the data, but the data isn't the real key. So let me move to the next piece of it, because you can consume it as a data API. That's the most basic foundational, but more important is consuming it embedded in solutions. And so we use this foundation, this foundation of external data to power some of our analytic solutions around fan insights, around demand and market insights in the retail sector. We're using it in the insurance industry, we're using it in the utility industry, and give you a couple different examples. I'm going to start with a real simple one. If I told you it was 50 degrees, I'm going to use a weather example, you wouldn't know quite to do with that. If I told you it was 50 degrees in the middle of August in Miami, you would say people are going to be wearing puffer jackets. People are going to think it's freezing. If I told you it was 50 degrees in the middle of February in Boston, people are going to be out there in shorts. So understanding the context really makes a difference about what does it mean that it made 50 degrees. So a large part of what we focus on is helping people make sense out of the data, not just get the data, but understand what to do with it, how to interpret it. And part of the key focus we have, and it's one of the key messages I've given my team from the beginning, the focus is to give people actionable insights with a real bias towards action. And it comes in several flavors. The first flavor is we bring in all of this data from all these sources. What we also give people is tools to figure out which data is relevant for a given use case. Don't get lost in the sea of data. Find out what data is useful in a given use case. Help people narrow the focus down to just those pieces that are relevant. That's number one. Number two is we focus our data science analysis of the data behind the signal and noise. There's a lot of noisy data out there, but there are signals embedded in them. In fact, I talk about data coming in two different flavors. There's contextual data, like census data or bureau labor statistics. Interesting, useful, you need to have it. It shows up in all kinds of data models, but it's foundational. Then there's data that has embedded signals. Twitter has got embedded signals. It's rich with embedded signals. The weather company data is rich with embedded signals. If you can pull those signals out, every one of those signals is an opportunity for a business to take action. Signals are what trigger action. When you have an action-oriented mindset, you try and hunt for signals. So you're not just delivering raw data? Correct. Delivering raw data. How do you package that? What's the secret sauce behind it? The key packaging of that is in these solutions. That's the layer where you embed these capabilities into something. Think about an insurance industry solution where you're generating alerts around property damage, or generating alerts around hail storms that are going to affect automobiles, and you can send alerts out to policy holders for an insurance company and let them know what's in their path and let them know what to do to get around that path and how to avoid that path. So that's an example of the type of thing we can do. We're also doing work in the utility industry to help them forecast outages when you have storms that are coming. How do you know where the highest impacts are likely to be? How do you stage your crews and your repair equipment so that you can maximize your ability to respond? Similarly, doing similar kinds of things for cities around emergency management solutions, police, fire, safety kinds of solutions. All of those things can be guided with better insights about where issues are happening and how to respond. When you talk about understanding the world around us better, obviously weather plays squarely into that, Twitter plays squarely into that, but what other kinds of sources do you not have now that you'd like to acquire? So there's an almost endless list of companies who would like to partner with us. In fact, I've had to kind of push a stiff arm for a little while because we had to get our first offerings out the door. There are lots of companies who have data in the consumer space that we think are interesting. In fact, when I think about the data areas of interest to us, I put them into a couple of different categories. There's consumer data. Think about the Nielsen's, the IRIs, the Equifax, the Experience, the Axioms. All of those are companies with interesting and useful consumer data that could be valuable to some of our use cases. There's also data around geospatial, and geospatial is a broad interest area, a broad category. We're working and talking to lots of companies. We've got mapping solutions. We've got satellite imagery solutions who are bringing in lots of geospatial data. We're also interested in various Internet of Things use cases. Almost every Internet of Things use case has a major data element to it. And so you want to have data around cars if you're working in the IoT or telematic space. And so there's a range of different IoT focused areas we want to bring in data. Now, are you going to guide your customers toward processing this data in the cloud? It seems like this is the focus of this particular announcement. Is the cloud really the best place to process Internet of Things data? So the cloud is often the best starting point for it, but process happens in lots of places. And so part of what we're doing in the cloud is helping bring the data in, do this alignment around geospatial alignment around the people, places, and things. And then once you've got that done, whether you do the next step of processing in the cloud or in your environment is a function of the use case you're trying to go do. Many customers will choose to do that last step in their internal environments. Others will choose to do it in the cloud. It's a function of what you want to go do. But we're giving you the ability to integrate the world in the cloud because that's the best place to do that. Well, you're making Bluemix sort of the integration point initially. The APIs will all be available through Bluemix. Is this a way, how does this translate into sales of your analytics products? So people use our analytics products for lots of purposes. Some of the core technologies that are embedded in what we've built here comes from our analytics portfolio. So our ability to do this entity resolution actually draws from a lot of the same underpinnings as Watson. So if you've seen the slide that the Watson team shows where they show all of the underlying API capabilities that are part of the inner guts of Watson, we use many of those same technologies in our system. So when we talk again about understanding the world and reasoning about the world, we're the understanding piece and we are using some of the same underlying technology. So you've got Watson technology components, pieces of the analytics portfolio and you've built specific software around that and then your business model is what you license data from these sources like Twitter and then you are allowed to resell that as part of that and so, I don't know if you saw Joel our intro, we were just talking about IBM's opportunity to take industry specific data and package applications essentially is what you're doing. It's a service. Now, that's a differentiator and a weapon. That industry knowledge comes from years and years and years of consulting expertise. How did you, that could not have been trivial to say, okay, we have this I had mentioned, former PwC 15 years ago, one of the best acquisitions in the history of the computer industry I've called it the second best behind VMware. Give me that. Maybe, that's why you acquire them. But how did that happen that you were able to extract that knowledge I mean, organizationally and just the process of doing that could not have been trivial. That's a great question. It's actually one of the observations that made to the analysts last night when it was briefing analysts. We put all these pieces together as you alluded to your opening, formally organizationally in a tight integrated unit this year and we've been moving that way for a couple of years. We really formalized the organization structure this year. I will tell you, we have a better, closer tighter working partnership now with our GBS counterparts than I've seen. I've been in the company almost 35 years. I've been through all the ups and downs. We're really working well together. There's a project that's going on right now that I'm not sure I can name the exact customer, but it's a project that GBS has been doing and this particular customer is in the CPG industry and they realized that 80% of their business was dependent on volumes coming from about 20 30 cities around the world. It's a very specific to key metropolitan areas. And so what we started doing, the GBS team started doing is building integrated information services around those cities. All of the data around those cities. Everything you can imagine in terms of the mapping, the geospatial data, the economic data, the activities, the traffic data, everything around those cities. And what we're realizing is that that unique city asset is one that I can put through my service as a way of selling you, not just general insights of service, but I can tell you a lot about New York City. I can tell you a lot about the economics of what's happening in New York City. If I told you how much net worth was walking down Fifth Avenue at a given point in time, what would you do with that? If I told you how much spending power is going down a certain road in the burrows, would you put a store there or not put a store there? And it starts to get pretty cool the things you can do when you focus on specific regions like that that have got a lot of economic weight to them. So the interesting observation here is that for decades now IBM Global Services has taken technologies that IBM has invented or acquired or whatever and then built services for the customer. What do you need? We can build that. We'll just grab from the grab bag and build that. You're flipping that around. You're saying, okay, let's take the IP and the knowledge that you have, put it into software and scale it. Exactly what you're doing. Another example in that space is we've had years of experience working with retailers across every SKU you can imagine and in the process we've learned what are the factors that drive demand for those SKUs? Not from one company but from hundreds of companies around the world and so now we know what factors to build into a model that will do demand prediction for that SKU. That's IP assets developed by our services team that we can harden into insight services offerings that I'm building. Will you go out on behalf of your customers and search for information sources, actually strike licensing deals with specific. If I come to you and say I have a traffic format, I have a logistics application, I want Waze data, will you go out there and try to strike that deal with Waze? So that's a very germane question. As you might imagine here at Insights, I've probably done two dozen client discussions in the last 24 hours and I think I get that question in every meeting because everyone looks at that and has a set of issues with that and I approach that in a couple different ways. First of all, our GBS team would say yes to that. In fact, almost every engagement they do with clients where they need to do that, that's what they do just as a matter of course. So that's an ongoing process that that team does. But what customers are actually asking for is a little bit more. It's closer to what you just said, can you go work out the licensing details, put in place a contract, put in place a sourcing, put it into your service so it's easy for me to consume, do all the rest of the stuff and I would love to do some of that for our customers. This is just a matter of how much bandwidth can you do and how much of that can you do with all the other applications around that. So I look at that as a series of requirements right now that is helping me prioritize who do I go after as opposed to being a specific service. Now listen to all those requests though, literally every day and use it to figure out where to go. What's the history of your business unit? When was it formed? How long has it been around? Talk about the market traction. Yeah, so this is another sort of just aspect of my personal history. I've been in the company as I said for almost 35 years. I spent 20 years nurturing new businesses inside the company. Getting them off the round, getting them formed, getting them started. So I was asked to go build this business by Bob Picciano and Jenny Rometti about a year ago. So we've been working for a year to get the basic service launch. The first official launch of the service was yesterday. So we are just coming out the door. Okay, so from a revenue standpoint it's not meaningful yet. When do you expect it to be meaningful? How long will that take? Is it 12 to 18 months before it really starts to hold? It will take a little bit of time. It will take some time before the annuity is really material at the IVM level. What is reasonably more material in a shorter time frame is the number of GBS engagements coming out of all this activity. We're doing a lot of business with Twitter and we're getting a lot of engagements with clients around Twitter and most of that is pulling through a little bit of my service business but I'm a little tiny piece of most of those deals right now. We're sending around the GBS side and that ramp is building nicely so we'll start to see more impact from that over the next few years. Twitter is obviously going through some management changes but as a developer, we're a small developer, Twitter wasn't so friendly to developers and new management is sending a message that we're going to be much more friendly to developers. IVM is reaching out to developers so there's some opportunities there. Definitely true and Twitter as I said is a data source that is rich with signals but learning how to extract the signals from that source has not been easy. Because it's a sewer. There's a lot of noise in those signals and finding the signals and extracting the noise and making them actionable is actually a lot of value. I recall discussions having with a particular data company this is about six months ago and I was talking about the new business we're building and when we did the opening introductions he kind of looked at me and said after about an hour of discussion he said I've decided you're going to be my most important partner because the value we were helping him do is make some of these sources more usable, more consumable more directly embeddable into the services he wanted to offer the marketplace and that's a lot of what we're doing. You're applying technology to the data analysis to the data and the data like you say these guys are partners you'll buy their data help them go to market with their data. We're really not in the data business we're in the insight business we're pulling together multiple data sources to find new instances. And you need data as raw material to do that. And we end up being a route to market for so many of those companies we help enhance the value of the assets they own. We've heard a lot about Twitter in the last two days we haven't heard a word about another big partnership you did this year with Facebook and Facebook not only the data from Facebook but the data from Facebook's holding such as Instagram and WhatsApp could have significant value to marketers where does that deal stand? So the Facebook partnership is still in its early stages it was actually originated in our commerce organization they've made good progress on coordinating with Facebook on ad placements and the use of Facebook as an augmentation of the full advertising offerings for some of our commerce clients the effective use of the data that can potentially come from Facebook is an area that we're just beginning to explore still very early for us on that. So it's primarily an advertising optimization vehicle right now. Exactly, it's in the commerce arena and that's the primary performance. Well Twitter is just interesting I mean it's the real-time data source it is the news data source and the moments is just now there's new innovation that allows you to... So we had Chris Moody on stage yesterday in one of my sessions and he has this little video he plays and it's a spinning globe with little lights flickering for where there are treats going on and the globe spins every metropolitan area is lit up with these lights and he asks people in the audience so what do you think this is? And people think well it must be like World Cup or something it's some big global event that causes all this and he said no, no, no, this is chats on Twitter on a random Tuesday evening about snacks. So, you know, there is all kinds of stuff that's out there and the trick is figuring out how to pull it out in ways that are useful. So another little thing that's helpful to actually see this we have an offering that's being made available here in the fourth quarter from my partners in the solution side from Alastair Renny's team that's around fan insights. So this is in the sports arena where we provide insights on fans and what fans like, what fans don't like it's useful for merchandising, it's useful for working on parking issues and stadiums and working on the stuff that's sold in the concession stand. So it's useful for all those types of use cases. What we found in working with people in the sports industry or people in the media entertainment industry is there's not a lot of good data on consumers for them. You know, the Nielsen's and the Axioms and the Experience provide tons of data to the retailers about a universe of data around their consumers to draw from. But if you're in the media entertainment industry or the sports industry you don't have nearly that sort of richness of data and Twitter's turning out to be one of the better sources to get good insights about what people like and don't like to a certain source. We had Nate Silver on in 2013 the famous statistician 538 blog and so forth predictor of elections etc and we asked him at the time what he thought about Twitter as a data source and he said the data is not good enough it's too early he was not very optimistic about it. Maybe what you're finding is that it's really hard the data is there. Now, but you may be using non-conventional techniques to analyze that data. What are those techniques? Is it just sort of you know, text analytics? Is it some of the NLP stuff from Watson? It's not traditional stats. It's not R. It's not SBSS. Well, you may use some of those underlying tools but just having those tools doesn't give you the answer that's for sure. So when you talk to people who have a lot of experience in working with Twitter data there's a fairly consistent pattern. It's the same pattern we've found but others have found it too and it's true across a lot of the social sources part of this is understanding what voices do you listen to what voices are just not useful so how do you filter out which sort of Twitter handles are worth following and from within that which sort of data sources which elements of the data set are worth pulling out and how do you use them people think of the tweet as just being the tweet but there's actually I think it's 65 different data fields associated with each tweet that you can analyze in different ways to find different patterns. Finding ways of clustering those around common either personality profiles or segmentation profiles or interest profiles is a big first step in getting to the voice you want to pay attention to and then extracting using sentiment analytics and other processing techniques extracting what they mean by the comments they're making is another level of value. Essentially building a psychographic of the true influencers not necessarily the guys that are just spamming the hashtag is clout in your opinion a good indicator or is it too superficial? It has not been a really good indicator to be honest it hasn't been. It's falling away now. It's nice and it's a fun little metric but for analytics it never really gave you the insights that you needed. They wanted it to do that and it never really did and similarly these network analyses of just how many people listen to other people that's also proving to be not as effective as people thought it was going to be originally. So there's a couple other layers of approach you need to get into before you find it. You do have an election coming up in about a year it seems like an ideal opportunity for IBM to show off it's analytics expertise not something typically IBM has done though to get into forecasting to get into politics is that an option this time? Well it's not one that I would personally pursue I've got too many other things I should have do but I wouldn't be surprised if there weren't people at IBM who were wondering about exactly that idea and some of my friends and Watson were giving all the things that they explore and they explore are amazing sets of things. I think they might have fun with some of that. Donald Trump influencer or troll? I'm going to do you guys to answer that question. Well Joel really fascinating discussion congratulations on incubating this new business unit really exciting thanks for watching and wish you the best. Great thanks very much guys. I appreciate it coming on. Alright keep right there everybody. We'll be back right after this word. Check out ibmgo.com for the social experience at IBM Insights. This is theCUBE we're up right back. Great.