 Live from Boston, Massachusetts, it's theCUBE at the HP Vertica Big Data Conference 2014. Brought to you by HP with your hosts, John Furrier and Dave Vellante. Welcome back to Boston everybody. This is Dave Vellante with Jeff Kelly. We're here at the Seaport Hotel, the Weston in Boston on the waterfront. This is the HP Big Data User Conference. The hashtag that we have on the crowd chat, crowdchat.net slash hpbigdata2014, so there's a crowd chat going live. And we have been unpacking a lot of what's going on in the big data world, a lot of action with analytics. In particular, we're talking to a lot of customers. Today really is the big customer day. Craig Snodgrass is here as the Chief Strategy Officer at Cardlytics. Cardlytics is all about reaching everyone. Exactly. Thanks very much for coming on theCUBE. Hey, no problem. So tell people about Cardlytics. Very interesting firm and obviously data driven. Yep. So we exist because of big data, right? If it wasn't, you know, everybody talks about big data like we exist because big data is around. Our big data is access to payment data, purchase data. The simple act of swiping your card generates just an immense amount of insights. If you think about that, right, when you swipe your card, you know where you're shopping, what data it is, how much you're spending. And that is our big data. So we take that data and we use that data to help advertisers bring in more and new customers. We partner with banks, you know, your data stays at the bank. We have no access to personal information. And we can look at customer day to purchase data and say, wow, we know somebody who's shopping the category but not shopping with you. Or we know somebody who is shopping with you. And there's still another three or four trips that you can get out of it because they're a heavy category shopper. We know how to attract those people in. We can target them and then through our partnership with the bank, we actually offer these customers rewards through their bank to come visit an advertiser. So you'll get 10% cashback, 15% cashback when you go visit the advertiser. No coupons are needed. Nothing needs to be printed out. You simply swipe your card and then the money gets deposited into your account. So it's all done electronically. It's all done electronically. How do you notify the consumer that, how do you get them, you know, to act? Yeah, there's a couple ways. So, you know, in the past 15 years, you know, everybody's obviously moving, moving online. Banking has moved online in a very heavy way. People log into their bank account nine times a month. And that is where we actually get in front of the customer and give these, you know, the advertisement. We meet them on the channels that they're already at. We don't have to create new channels. And then with the mobile app, they check the mobile app, you know, 15, 20 times a month. We're on the mobile app as well. So if you open up your app, you'll see, you know, each bank, we're 400, you know, approaching 400 banks. And each one has a mobile app. And you can actually see all of the, all of your offers inside of there. So we get them, we get them online and mobile. So you're not profiling me. You are going down, as Colin said, you're going down to a segment of one. My persona. That's our reach everyone. It's three words, not reach everyone. You know, so that's, it is truly, we are looking at a customer and it's just with the purchase data. You know, where is it you like to shop? And therefore we are trying to find offers that match where it is that you like to shop. Now, the obvious question is where are we in terms of the state of privacy laws are in their infancies? Are you guys self-policing? Are there maybe, you know, laws that are coming down that I don't know about? Maybe you could talk about how you protect privacy. Yeah, it's a great question, right? And I actually do think, I think it's in its infancy from really having established rules. But from a policing standpoint, the way our technology is set up is no, we have no PII. Like we have no access to it. It doesn't get into any of our, it doesn't even touch our software. The banks of course have all the PII, but our software doesn't have any access to that. And then we never share information with the advertisers at an individual level, right? It's only aggregate reporting and everything's lumped together. So from that standpoint, the self-policing is probably, I don't know if that's the exact right word, but we don't have any access to anything. And then nor do we ever give anything at an individual level out. So really from a protecting the customer standpoint, there's nothing that can really happen with the data at the customer level. And that's what we want, right? We can get all our insights. We can aggregate everything up, but nothing ever gets shared back or touched at the individual customer level. Well Craig, it's an interesting business model and you mentioned, you know, you're never sharing data with the merchants. Whether it's a retail organization or maybe another financial services firm that wants to get you to do some business with them. But at the same time, that type of data, I imagine your customers, those merchants would love to get their hands on it. Oh, absolutely. I mean, I'm sure they'd love to bring some of these capabilities in-house. How do you provide value to merchants and also kind of guard that crown jewel and keep the business model crisp like that? It's interesting you bring it up because when you look at advertisers, they have this phrase called the last mile. And that's what it is that they're trying to measure. I know that you saw a TV commercial. They're finally now able to measure down to, it's called addressable TV. They know what it is that you're watching, right? And now they want to know, okay, so when you saw that commercial, did you go into the store and make a purchase? A lot of people have online data, but, you know, online purchases only make up 7% of what it is that we purchase, right? It is a small number compared to there. So the last mile is what everybody wants. This data is the last mile, right? So once people understand what it is that we have, they instantly go to, well, can we just buy the data from you? And the answer is no. You're never going to be able to buy this data, right? You have to work through us. We're protecting the customer, you know, come to you. It is, we're protecting the data through our business model. And so we're not giving that data out. But what we can give out is insights. So if they want to understand, you know, in a geography what's happening, we can aggregate the answers back and say, here's the share of wallet that you have in a geography. Or if you have, like, your top decile of customers, your top best 10% of your customers, here's where else they shop. So it really is at an aggregate level. They can't tell at an individual level what's going on, but they can still get these powerful insights. And they can get insights from as simple as, wow, we just had one advertiser purchase a competitor, and they have stores that are nearby each other. And the question is, which of these two stores should I close? And you can actually look at, you know, which one has a more loyal customer base. Because when you close a store, your competitor will get a bump from that, right? So you've got to choose wisely which of those stores they do. So typically in the past, they would just kind of look at, you know, what volumes are there. We can actually look at them and say, hey, the customers that shop these stores, where else are they shopping? And is there one that has a more loyal base than another? And if so, you don't want to close that one. And the same thing is, where should you be building a new store? Right? Because now we do, here's an underserved area. You have people that are either driving a long way, or you have an area that has people that look just like your customers, but they're not there today. If you were to go build here, this is what you would expect from a volume standpoint. So you guys always have a lot of background in the credit card business? Yeah, debit's been the real thing for a while. I know a lot of people bring up credit. In the past 15 years, the move to debit has really been the thing that is making this model successful. So 15 years ago, it was only about 20, 30% of all purchases were made in store with a card. It's now 70%. And that growth has come from debit. Debit really has gotten rid of cash. Debit has gotten rid of checks. You know, when was the last time you wrote a check in the store, right? But, you know, America is becoming a debit nation. So these credit cards are for travel and large purchases. And then debit cards are for everything else. We're catching up to Europe. Yeah, yeah, yeah, exactly. France has been the debit nation for quite a while. So you guys started at the end of last decade. Right. It's almost been like a perfect storm for you. Because in 2008, you're talking about the financial crisis just hitting. So it was kind of an interesting time. But then technology has advanced so much since you guys launched the company. So I wonder if you could take us back and take us through that journey. I mean, the vision of what you guys, you know, wanted to do and the technological capabilities have advanced so fast. How has that enabled your ascendancy? Yeah, yeah, the technology journey has been quite the ride. The easiest way to say it is, when we first started, we had mid-tier banks. And the technology that we're using worked just fine. Then we landed a large bank and then the technology was not going to keep up. So there was a lot of redesign that we had to do in the architecture to begin with. But from an analysis standpoint, you know, we were using, you know, a competitor to Vertica at the time. And to get the insights that we wanted, we couldn't. Was it a traditional enterprise data warehouse? It was a traditional, it was a very famous name that everybody will know in one way, shape, or form, right? Everybody's used it. And it, we couldn't keep up. What happens is, once you sit down with an advertiser, and this is a bit off-topic from a technology, but it will come back, right? Once we sit down with an advertiser, in the beginning of the conversation, they kind of understand what kind of data we have, but by the end of the conversation, they're like, oh, you not only know what they do when they come in my store, you know what they do when they leave. Like, you have a full 360-degree view of my customers. It's like, yes. And then the fire hose is open with pent-up years of questions that they want to ask. And that's what you were supposed to get from your traditional enterprise data warehouse, but you couldn't because it took too long, and it was too expensive, too complicated. Way too long to get it back right, exactly. So we moved to Vertica, and what we were looking for was, our quest at that time was rapid cycle insights, because you would go meet with an advertiser, and they would be excited. They're like, this is what you can answer? What about this? What about this? What about this? Well, it would take us too long to get back to them. At that point, they're less excited because they never got the answer. I mean, we're able to effectively just flip it around same day next day and just give them the insights. I mean, Vertica allows that. Like, you'll come up, and when I'm driving, I'll come up with a question, and it's like, I know I can just sit down and write a quick script, and I will get an answer back, and that will lead me to what it is I really should be asking. Like, that's the rapid cycle insights. I get something back rather than waiting forever, and it's like, oh, these are the three questions I should be asking now that I know that direction. Is that a function, the ability to do that rapid cycle iteration, the function of the ability to ingest data rapidly? It is not the ingestion of the data. It's when you sit down in front of your console, it's instant feedback what you're getting back. You don't have to take samples. You don't have to ask a small question and try to project up. You use the entire full data set, and your answer is back in seconds. We had scripts before, they would just never complete. Sampling is dead in your work. Sampling is dead, and most important to me is, I believe, surveys are dead. So what these advertisers before would have to go and survey, and to me, every one of these swipes is a survey. Like, you know, we have, you know, billions and billions of surveys going on at all times, so then you can start asking, you know, something with survey is you have to ask your exact question and make your survey exactly that. Yeah, I wish I asked that question differently. And then six months later and, you know, hundreds of thousands of dollars later, you finally get your answer on that one. With the data, you know, you can just quickly pull it back and give it. So I think it's, I believe, yeah, sampling I think is on its way out. I am hoping surveys are on their way out as well. You really want to measure them. So thinking about your role as the Chief Strategy Officer, you've got the demand piece from the advertisers, you've got your data sources, you've got the technology piece, you've got an ecosystem, there's probably other components. Can you sort of paint a picture of sort of how you spend your time when you're kicking back? I will tell you how I spend my time. The best way, for right now, the way I'm spending my time is that I am now on a new quest. Vertica has fulfilled our first quest, which is the Rapid Cycle Insights. We can now bring in the data and answer the questions in a fast manner. It's to use a sequel. You don't have to go train, you know, on a highly specialized skill set from there. Now my quest is I need the simplest path forward to productize the insights. We're not going to be able to write sequel all the time for everything. Everybody wants this, right? I want a BI tool that sits nicely right on top of the data and allows people to explore and can package things up and still protect the individual level of data but still give the insights. That is truly the quest that I'm on now. I am reading articles and I am looking at anything. It really is looking at what other folks are doing on the BI front, which is why I'm so interested in what Vertica is doing because, you know, before we would cube everything up and then have a BI tool connect to the cube. We don't even have to cube things up now. We just basically hit the data so fast we hit it right away but it's still a little bit clunky right now and how we have set it up. So my quest is and what I'm reading is what is the easiest, slickest, most beautiful way to productize our data? So that's a scale challenge, right? It's a scale and ease of use, whatever you want to call it. So right now it's complexity is the barrier to that. You got it. Now what's involved in that? Obviously visualization is a component of that. You're describing sort of a self-service BI, putting intelligence in the hands of business users. I liken it to Google, right? Well, just to sit down in front of Google and enter in your question and then you will quickly get your answer. That's what we want. Google for BI. Google for BI. And it's so, there are so many questions. You can't productize everything. Just like there are so many websites. There's no way that you could build a great taxonomy for what you want to go search for. So therefore they go search everything and they're trying to understand what it is you want to learn and then they're packaging that up and feeding that back. That's what we want. It won't look like that. That'd be great if it did look like that. Maybe we shouldn't even call it BI. Right. Right. Maybe we need a new name. Well, yeah, I think the term BI business intelligence has a lot of baggage to it over the years. So that's not a bad idea. Maybe we should think about that. It's a new term. But just to dig into that a little bit more. So when you're talking about essentially a self-service layer, you want to roll this out to your internal customers or your external customers or both? Well, it's definitely on the internal side. The external piece is one of the benefits of working with us is that you get insights. Part of the advertising is of course you get the measurement, you know, who saw it. Again, not at the individual level, right? But we track who saw it, who clicked it, and whether or not they came in the store. So we truly have the measurement effectiveness. We've got that down. But it's the insights, right? Because we don't know what their analysts have to go figure out. We don't know what they're up against. But we know this data can help. So we can't just create prepackaged things because the questions are all over the place. Getting that tool in their hands in a safe and yet usable way for them to go grow their business as well. Because I'm sure they want to take your insights, merge that with some data they've got internally, and let their analysts go at it and come up with some other types of inferences and insights, maybe not related to advertising directly, maybe related to something else completely. You got it. Right, store effectiveness. There's so many things that they want to understand and we can help with that. But I don't want to actually build a tool for every one of those questions that come up. We can't be able to keep up. So we actually have an insights team who works individually with our advertisers to go answer those questions. And again, the questions are not just industry specific, but you're right. An individual advertiser might be facing something right now that this data could either be 100% the answer or help aid the answer better. And that's really through that conversation that we had. It could be something like product development. We want to pursue a new product based on any number of data sources you want to analyze and some of that conclusion. One of those, of course, could be the data that you're analyzing and the insights that you've come up with. One of the tools that we have is this interaction index. What it means is, if you're a Macy's shopper, who do you over-index against? So Macy's shopper most likely also goes to and now I'm going to make it up Nordstroms or whatever. But that's what you'd want to know from a product line standpoint. Oh, our existing customers also like to shop over there. What is it that I can do to start bringing some of that business and have it all happen in here? That's really it. One of the ones that we had done before was David's Bridal for wedding dresses. And it over-indexes against Chick-fil-A. I don't think David's Bridal is going to go do anything to start bringing the Chick-fil-A's over. But it's just one of those... That's interesting. Is that a correlation? First conversation? It's a beer and diapers thing. Well, and then the other one is people who purchase taxed services and it could be turbo taxed but it could also be fractional are way over-indexed on bars. Right? I mean, it's just one of the... I don't want to go through causation. I want to go through cause. Like the data is there. There is a very, very strong correlation between the two. And how about data sources? Can you talk about that a little bit? I mean, is it predominantly cards? Is it exclusively? I mean, what kind of other data? How diverse is the data set? The data sources really focus on purchase behavior from the banks. So it's debit card, credit card, bill pay, ACH, those types of purchases. We do not... When I go out and read about big data and the volume velocity and all that, we're not capturing a whole bunch of other unstructured data. We really do have a... We are drowning in signals. It's whole separating the signal from the noise. We have just a fantastic pipeline. Yeah, exactly. It is taking all those signals and answering all the different questions. So we have a structured data source. That really is the main driver for all of our data. So not huge amounts of data. It is not... When people talk about the petabytes, we are nowhere near that range. And no real need to change that in your strategy in the near term anyway. I agree. There's an ISO 85, 83 standard that has all the payment card data into it. It's very, very structured. It gives you exactly what you need. I'll take it all day long. Well, it's about the value you can extract from data, not the volume. Yeah, and I think that's... Back to the signal and the noise, right? We are... We have enough signals. One day when we start getting new data, that will start adding the noise and that's probably where the volume is going to start to ramp up. But today, you know, our big data really is how do you take this data and quickly manipulate it to go get your data? I mean, you think about social streams and Facebook and Twitter. It can't compete with the data source that you have. It is today anyway. It is. There's no better sentiment than going back to your store. You go to a store one time and you don't come back. Right? I mean, it's... They've answered their sentiment right there without having to hashtag, tweet or do anything from that. Considering your customers, I wonder if you could weigh in on this topic of the role the CMO in marketing relative to IT and how CMOs are getting more involved in obviously analytics and being much more data driven. How have you seen that evolve over the last even just a couple of years? Yeah, I believe the CMO is a very, very tough position right now because I'll use a bank example. A bank has had years to get used to the data volumes that they have because they've always been saving it. Their underwriters have always wanted it and all that. Their curve is slow and gradual but still pretty hefty amount of volume. A CMO has gone from zero to massive scale in three, four years and now the finance, the CFO is asking for proof that the advertising is working. The CMO is in this position of I've got to go get new customers and now I've got to have all this data I've got to go through. A large budget that now is being scrutinized exactly how I'm using and all that's being pushed on them in a short amount of time and then there, if you look at there's a trend of CIOs becoming CMOs. I don't know if there's a trend of CMOs becoming CIOs but it's becoming much more technical now because all of these questions can be answered with data and you know they're probably pitched we talked to CMOs right they're probably being pitched all the time about new data sources how it can help them out, what it can do how they can separate the signal from the noise how they can measure effectiveness all of those components are all being dumped on them and they haven't had 25 years to get ready for it it has happened recently. It's a tough position. How are they adapting? Are they hiring new people with different skill sets? They are hiring new people and it's interesting when when we do talk to CMOs and you go meet the analysts and they're on their side they're really good you kind of come in a little bit cocky like oh they know their business cold they know their business cold and they know the technology and they know all that they're much more nimble about trying something new than somewhere that's had 20 years of something that's been sitting right it's like starting a bank after 2008 you don't have any toxic debt you just get to start the races have come down and all that and that's a little bit where they are as well they're able to start not necessarily from scratch but they're able just to go pick up something brand new because they don't have a ton of legacy things that they have to worry about or when they do get some access to new data it can come with the technology that's with it as well So you see them doing the old n-run around IT and going to cloud services I think it's a mixed bag I think there's two examples I've seen I definitely have seen the rogue you know we just need I need to go find out if we should even do this first so let me do a simple little proof of concept myself then I'm going to go to the technology team and say this is what we want to go do I have seen that more and more I think what we're seeing is the CIO and the CMO are just becoming tighter like the CIO now realizes oh I can really help out the CMO and those conversations are maybe replace the CMO that has happened several times that is the conversations that are going on now you'll see when there are when you come to conferences like this there will be some folks from marketing and there will be some folks from the technology group here and they're evaluating it for different components and really if you think about on the technology front they're dying to be able to go provide this kind of business value that can propel the business moving forward like that's what everybody in technology wants to go do and they realize wow we have the skill sets we have the ability to go really help out the marketing department let's go figure out how to make that happen yeah I think that's a really good point big data is allowing IT and CIOs to actually start adding more business value and being less just about keeping the lights on and systems running I have a feeling if there was a social graph of departments that you know ten years ago the marketing department and the technology department probably did not have a strong connection between the two I have a feeling that social graph is dramatically changing now and if you start thinking about wow like we can really help out here it's probably becoming even stronger than some of the other relationships that are out there how do you guys deliver deploy is it a SaaS solution? it is our technology for the banks the bank buys the hardware and puts our software on it behind the wall so I know software as a service is the best answer and then of course it's distributed right we have multiple banks we're almost a 400 banks so we have all these installations everywhere and then there's a stream of data that comes back to us in a centralized point so we have to create campaigns for an advertiser we create a campaign in our homegrown software and then it publishes it out to all the different installations that we have and then it organizes the information back into the warehouse so that we can then use that for analytics and reporting but the advertiser is paying you obviously how does the advertiser get access to the data? today they have a portal a gateway that they can come in and they can look at how their campaign is running and the key is they get your standard impressions and how many customers are talking to and all that but they actually get to see the purchases that these impressions are driving so we have a, you can watch people who saw this impression drove this many trips and they can see that real time and that's not something that they can get with their other reporting they just send standard impressions and clicks and they pay by program they pay by they pay by people coming into the store it is a pure performance driven model if we run an ad and no one comes into your store you don't have to pay we have to be able to drive that and of course because we have access to data we can do test versus control and show the incrementality that the advertising had as well so they can truly measure effectiveness, return on their ad spend so you baseline it and you agree on a baseline with the customer and then from there it's essentially charging for the incremental revenue that you're driving we guarantee an incremental revenue we don't just charge for the incremental we charge for all of it but it all goes into the rate it's the same thing it's not working they know what the ROI is that they're going to get that's awesome all right Craig great story we have to leave it there thanks very much for coming on theCUBE congratulations on getting the business off the ground and look forward to watching you guys in the future I appreciate it thank you very much all right keep it right there everybody this is theCUBE we'll be right back after this word