 Live from San Francisco, California, The Cube, covering MarkLogic World 2015. Brought to you by MarkLogic. Now, here are your hosts, Jeff Frick and Jeff Kelly. Welcome back everybody, this is Jeff Kelly. We are live on The Cube at MarkLogic World 2015. Here in Silicon Valley, I'm joined by my co-host for the rest of the day, Jeff Frick. Jeff, thanks for joining us. John had to head up to San Francisco for some very important business, so I'm happy to have Jeff here with me for the rest of the day. We're here with our next guest, Mark Unak, Chief Technology Officer at Codified. Welcome to The Cube. Thank you guys for having me. Well, thanks so much for coming on. So, tell us a little bit about Codified. I understand your whole reason for being is to make it easier to find product content. Tell us a little bit about it. We're helping e-commerce make products findable. I think we've all had that experience where we've gone and entered in something that we want to find and we either can't find it or we get too much. And people being squirrels, they give up really easily if they can't find what they're looking for. So, the findability problem is, first of all, what we're trying to do. And then second of all, we're trying to help retailers expand their product catalog, get data to the market faster and really offer more product. Well, take us back to the origins of Codified. Tell us a little bit about your background and how you got into this point. Well, I used to be CTO for a really large software firm called Amdox. Amdox was the billing system for AT&T. Used to be 20% owned by AT&T. I was responsible for the Amdox side of building U-Verse. And what we provided to AT&T was a centralized product catalog for all of their wireless and wireline services. And as a result, I got into this product catalog type of working where we tried to disseminate product information around the whole organization. Common product catalog, common product definition. Codified was really a very interesting play in that they've had great relationships with customers. Grainger, Motion, A-Sardware, Home Depot, and been helping them get their product data fixed up. But really, they ran into a technology issue in that they could make better data, but the technology couldn't take advantage of it. So I came and joined the team about a year and a half ago, and we built two new products that could really take advantage of better data and make it usable by these e-commerce sites. So walk us through that. I mean, why has it been so difficult traditionally to find content on a retailer site? I've been through that experience. You just find too much. You find too little. Or in my case, I think I might be missing something. Well, here's the problem for the relational versus XML world. In a relational world, you're trying to find the commonalities. And really, when you look at product data, look at a screwdriver versus a bathtub. The only thing that's common is brand price and description. All the other parts of what the description, all the attributes of those products are completely different. And so what we needed to do was go to an XML model where we could have unlimited attribution and separate attribution for every product. A relational database we would be doing hoops to get to that kind of model. So an XML database was really essential to making that work. And second of all, most of the databases are the search engines that are out there right now, specifically like the one that Oracle has and DECA, can only have 48 global attributes to be searched on. So since they're global, they have to be common across all the products. If I'm looking for a cast iron bathtub that can hold 100 gallons of water, how does that have anything to do with a bath towel? You know, it really doesn't. And how do you separate those types of searches and make them meaningful? One of the things we always show our customers is how bad their websites are. The other day I was at a meeting with CDW, 12 billion dollar company, 3 billion on their website. Their bread and butter is selling laptop computers. We sat there with them and we entered in 17 inch INCH, laptop computer, and zero computers came up. We typed in 17, zero. When we typed in 17 inch laptop. How many SKUs do they have in their real inventory? 10,000. We typed in 17 inch laptop, just took the computer off, and we got backpacks and a privacy filter. So that's the state of the search engines today. You can use a lot of examples. Another one, we went to a retailer, typed in classic fit swimsuits, and we got men's suits. So talk about how you're actually solving that problem. What are the attributes of our product, for example, which builds your product partly on? What does it bring to the table that allows you to make that a more accurate search result? We've instantiated the metadata in the tag value pairs. So we understand what a product is through the metadata, and then each product can have an unlimited number of attribute tag value pairs for attributes. So we can understand in the context, if you search for one quarter inch slotted magnetic screwdrivers, those are all, we solve that looking at the metadata rather than looking through the SKUs. So let's say you have a million SKUs, Granger, for instance, has a million SKUs in their database, they have about 16,000 metadata attributes and products. If I can resolve and parse the item in the 16,000 and then go in and get the SKUs that I'm only interested in, then I have a much faster search and much more accurate. Now, are you using third party data as well as part of building the attribution for these products, or are you still just focused on kind of the defined product description that exists kind of in the classic catalog? That's where our company comes along. We've been in this game for 15 years. We helped them manage this data. Now, we bring the manufacturer's definitions in with our bridge product and then transform the manufacturer's formats into the distributor's formats. So that's part of that heterogeneous merging of unstructured data that goes on. That's the other product that we have. We call it the bridge product. Right now it takes about six months to get on a distributor's website. We've lowered that down to about two weeks and the companies that are using it are as components in England and some of their suppliers like Siemens and Schneider Electric are going through our bridge product to get the data up on the distributor's websites much faster. So have you done any kind of before and after analysis with some of your customers? So well, look, if we've increased or improved the search capabilities by X% that leads to X% more business. We think that it's increased by 5%. We have two customers using search right now and we see them really improving their revenue. Now it's going to take a lot of more time for us to get really good statistics but we see positive numbers already. So that's what we think. Now a 2% increase in sales or a 5% increase in sales can be a big deal for a company. Well sure, for a big number. You're starting with a big number, 5% of that is impressive. Especially at retail when you're dealing with razor thin margins and really the competition is just so intense. A 5% is actually quite a large number. That's right. And how are you measuring that in terms of abandoned carts or abandoned churches or abandoned shops? Yeah, it's abandoned sessions. So what we know is that on the average if a person interacts with a website with eight events, eight times and if they get to that point without finding what they want, they're gone, usually eight on average. And we find that we're actually finding the product within that eight number much faster. We're kind of like at the six and by that we're saying that we're keeping people in session. Our sessions are longer, people actually find products and with a longer session, we think that they're buying more. And then do you also tie into recommended alternatives and recommended partner fairs and you know try to increase the best? We have a new recommendation engine and it's not like the Amazon one where people who looked at Plasma TVs also looked at beanies. Right. The idea is to provide, to give the merchandiser the ability to define what things that they want, what are equivalents, functional equivalents. So let's say that you have a quarter inch tip screwdriver. You can use whatever attributes you want to say what the functional equivalent of the screwdriver is and offer alternatives or even up so. So in this case, we would find a screwdriver that you're looking at and offer a set that includes that functional equivalent. So with the process of recommending a functional equivalent, you can either offer them something that's your brand name, like your own generic or you can offer them something that has a more higher margin. And that's a business decision that your clients make. We're giving the distributor the ability to merchandise. Now, when you go into some of your newer clients and you see kind of the, you mentioned CDW, you see the environment that they've got now. What's it like in order to actually make the change? Or I mean, we hear a lot about, is it a rip and replace situation or is it a gradual improvement of existing capabilities? How do you go in and actually make it happen without completely disrupting the way they're doing things? So usually these large companies have both an internal and external search engine, one for their customer service staff and one for their external. We've been winning on the internal search engine, the one that supports the customer service. So that's where we've been implementing our stuff where we can basically at some of these places, they have print catalogs in front of them because the search engine can't find anything. So we've been replacing that internal search engine. We're hoping that when people get really comfortable with us that we can take them to the outside search engine. So tell us a little bit about your relationship with MarkLogic. You're here at MarkLogic World, there's several hundred people here, Jeff. Tell us a little bit about how you guys work with them and your partnership. We're a customer. It took us a whole two weeks to come to our agreement with them. I think I was, I still, people talk to me about being the fastest sales effort that they ever had. So we knew what we wanted. Most of our people have been around the block. We were looking for an XML database. This was the best XML database we could find. We had both search problems and what we call acid problems. Problems of doing large volumes of data conversions. So between the two, MarkLogic had fit those requirements and it was a marriage made in heaven for us at this point. So no accident. You're from Chicago. Sears from Chicago, the original catalog. Seagulls, Sears, Montgomery wards. You name it, Chicago had catalogs. Yeah, exactly. So is there any tie back to that from back in the day or just happens to be where you get some great talent? You know, I worked on one catalog project when I was a kid. We worked at the Sears catalog. The Sears catalog was still up and running when I first started out. They had giant databases. I think that what we knew going into this was that the size of the databases were going to be enormous. Yeah, interesting. So what's on your horizon that you can share with us? I mean, what are you looking to do with some of the capabilities? Not just MarkLogic, but generally, I'm sure you're always looking to help your clients in new ways. What's kind of top of mind for you and what are you looking forward to the next six, 12, 18 months? You know, I think that what we're bringing to the table now that we're actually can truly define product information, you know, everybody's kind of gone down the big data customer behavior type of area. We hear people talking about, oh, how the customer did this or the customer did that. And a lot of the cognose type of talk is around modeling the customer. But it's all been done in a vacuum without understanding what the customer was trying to buy or what features of the product that really was of interest. And I think what we bring to the analytics part of it is the broad product information that could then marry with the customer and really understand buying behavior. And what features, what products, the way to merchandise and how that really is going to work. Now you can really say, I'm in an aisle, you're in an electronic aisle in a store. I can show you what's around you in that aisle by giving you recommendations. And I can tell you specifically what these products are because you can find them. So now that data of why you made that buying decision is going to be more exposed. And we as merchandisers can then go back and try to influence your behavior. Do you think retailers generally are, understand the potential of all this new data? We've been hearing it in the media for a while. So I'm thinking they're starting to get it. But where do you see them in terms of maturity of understanding what's possible and actually taking steps to change their culture to take advantage of? Well, they've had such broad, they've been painting with a roller for the lot. When you go to a website, you see big banner ads and that's about all they can merchandise with. Now they can get with this product data and with these tools that we're providing, they can get down to very fine grained product attributes that they can say, we know that this is what's important. This is what's really being sold, what people are buying. And they can start promoting those actual product attributes. So this gives them fine grained ability to tune their product catalog. Very good. All right, well, we're just about out of time. So I want to give you the last word. I mean, what's your thoughts on the show here today? Kind of your relationship with Mark Logic? I mean, what are you taking away from this event? I'm really impressed with how many people and how many different things are going on. You know, there's a lot of people doing a lot of really fun computing. And it's always good to hear other stories and other use cases of how the product is being used. It opened my eyes, some of these things about how people are using the heterogeneous data. So I have issues with content management systems. I need to merge video. I need to merge pictures. I need to merge rich text. And I've heard other people talking about those same issues. It was an eye-opener for me. Fantastic, well, Mark Unaq from Codify. Thank you so much for joining us on theCUBE. We appreciate it. We'll see you back here next year at Mark Logic World. Thank you. Thanks for watching, guys. We'll be right back right after this with our next segment. We're live at Mark Logic World 2015.