We'll see how Guess can increasingly predict how to satisfy its shopping customers, and we'll specifically look at how Guess's IT organization came to grips with adopting and implementing a big data platform to bring more of a democratization of data and better access to its employees.
To learn more about how Guess has slashed the latency between data gathering and actionable insights, join Bruce Yen, Director of Business Intelligence at Guess, Inc. The discussion, which took place at the recent HP Discover 2013 Conference in Las Vegas, is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions. [Disclosure: HP is a sponsor of BriefingsDirect podcasts.]
Here are some excerpts:
Gardner: Tell me why just plain, old relational databases and legacy IT weren’t doing the job for you.
Yen: About three years ago we began searching for a new database platform. We were hitting a lot of performance bottlenecks on our data loads and performance. We also saw the competitive landscape out there with lot of our competitors embracing alternative solutions to their traditional database platforms.
Gardner: What sort of requirements did you have to get to where you wanted to be?
Yen: The first thing was performance. We needed to improve the query performance. A lot of our users were asking us to do a lot of queries with very low-level detail inventory, and it was very costly from a performance standpoint to be able to serve those queries up. Some queries wouldn't even come back.
Secondly, from a performance standpoint, we wanted to make sure that a lot of our East Coast stores would be able to receive the reports early in the morning, and we were having problems just serving those up on a daily basis on time.
The last part was to support any kind of innovative analytics, any kind of cutting-edge analytics. We knew that that platform really wasn't going to help us do any of that. So we needed to find a new solution.
Gardner: We know one of your popular and well-known products is your jeans, Guess Jeans, but there is more to it than that. Tell us a bit about the organization.
Yen: Guess has been around for more than 30 years now and we've grown from primarily a U.S. retailer into more of an international retailer.
If you look at the '80s, lot of people from the States remember us for the triangle on the jeans. We were primarily a wholesaler in the beginning. Now, we have over 1,600 stores worldwide, and about half of those are run by licensees. We sell a wide variety of lifestyle products, targeting primarily younger women in their late teens to early 30s.
Gardner: So it's critical to understand that market, and this is a dynamic market. People's tastes change and tastes are also, of course, different from area to area around the world.
What have you gotten as a result of using Vertica? Can you give me some of the key performance indicators that now demonstrate what you can do when you've got the right platform and the right data.
Yen: I like to look at it this way. First of all, it's foundational, the foundations for just baseline performance. Have we met those goals? With Vertica we have. We've been able to meet all of our service-level agreements (SLAs) and serve up the reports on time. Not only that, but now we're able to serve up the queries that we weren't able to do at all.
When you move aside from the foundational, the next steps are analytics, being able to apply analytics and go through our data to figure out how we can apply best practices to see how we can gain a competitive advantage. We've been able to take our transactional data and look at ways of taking the stored data and applying that into our e-commerce site to get better product recommendations for our e-com customers. That’s something that we couldn't have done with our existing system.
We have our customer relationship management (CRM) system. We have our loyalty segmentation for which we use Vertica to do all of the analytics and we feed that data back into our CRM system. With the data volume that we have, we could not have done that with our old system.
So it's opened up new doors, but not only from a foundational standpoint. We've been able to meet our daily needs, but we've been able to set ourselves up to be competitive in this area.
Gardner: And has being able to gain the speed and handle the complexity prompted you to then seek out additional data to put into your analytics, so in a sense of not feeling limited as to where you can go and what information you could bring to bear?
Yen: Definitely. We've been looking at different things lately. We've been looking at different types of data—loyalty data and customer data—that we get from our customers.
In being able to give our users a holistic 360-degree view of what's happening from that customer standpoint, Vertica has been very critical in keep pace and enabling us to do that.
Gardner: Of course, it's important to get more data, manage it, and perform what you need to do with it. It's also important to deliver it in a way that people can use and to get to what we mentioned earlier about that democratization. Tell me how you've been able to deliver this out to more people and in an interface and device fashion that they really want.
Yen: That’s a great point. Everyone talks about big data these days, but big data, if you can't serve it up to people, if they can't use it, and if there's not a pervasive use of the data, is really useless.
We're pretty innovative in what we do from a mobile standpoint. For the last two years we've had an iPad app that's powered by the Vertica back end. We have this iPad app that over a 100 merchants in North America and Europe use.
It's been able to take a lot of the data, a lot of the stores’ data, a lot of the selling information. It's allowed them to travel to the stores, be in meetings, or at home on the weekends, and they can look at the best-seller information. They can look at the sales and do it in a way that is actually fun.
It's not just a bunch of dashboards or reports that you open up and look at, but we've made it very interactive and we’ve created workflows in there. So that really draws the user into wanting to use that information and wanting to ask different questions.
Gardner: And for this combination of the power of the platform, the quality of the data, and this distribution capability, can you give us some metrics of business success? Where this has helped you? Do you have any concrete things you can point at and say it's really working and here is how?
Yen: We’ve looked at that in different ways. One of the initial points that we're analyzing in terms of return on investment (ROI), the easiest one is the amount of paper that’s being saved. You can count up the reams, how much they cost, and multiply that, and there is some significant saving there.
But that doesn’t really excite anyone. It's great that we've been able to save paper, but the argument is, well, you also had to buy new equipment. These iPads aren’t free and the mobile device management software and everything else that's associated to it is a new ecosystem. So there is a lot of new cost there.
The exciting thing is being able to see our users look at the data and make the decisions. Before, they would have to stop at a meeting and go back to their desks. That decision that takes an instant now used to drag on for two or three days, maybe even a week, and I've seen that in action.
I can't give you an actual dollar figure, but I've seen them make decisions to change the allocation of certain items as they are looking at that information. As I was training some of our executives or power users, I would see them pick up the phone and actually make decisions to impact the business. So I know that it definitely has done a good job there.
The exciting thing is it's kind of democratized this information and this data and demystified it to a point where everyone can access it and everyone wants to access it. I’ve never seen users get so excited about a platform or an app. We've got emails saying, "Can I please have this app. I saw one of my coworkers using it. Could I please?" Before, we were never asked that way.
It was always, "Can I get a copy of that report. No big deal if I get it now or later." But here, people really, really want to use it, and we could tell that we hit something.
Initially, we had to deal with just our internal IT folks being very skeptical. A lot of the claims, "30 to 300 to 400 times faster in performance," "you’re only going to need a quarter of a DBA," were the first two items where a lot of us were a little skeptical—myself included—but the performance has really proved itself.
Aside from that, we have to look at it more realistically. How do we implement a system like this? A lot of it has to do with changing the data loads and that, in and of itself, takes a lot of time. That's one of the things that's always going to take a lot longer than we thought, and it would be a lot more challenging than we had initially anticipated.
The one thing that I'm proud of is that our team was able to conquer all of these hurdles, and also we had a great partner in Vertica. They were there with us in the trenches, even though we were the first retailer and we had a different use case than all of the other previous clients and customers that they had.
We took a chance with them, they took a chance with us, and it worked out. We were able to prove that their software works on a multitude of different use cases. As a retailer, we have a lot of updates with our data. This was three years ago. Their clients then, a lot of the telcos and banks were just loading data, not really doing a lot of updates with it. They were doing a lot of queries with it and it was coming back fast, but not really transforming the data all that much. So we had a lot more use cases like that and they were able to come through for us.
Gardner: What about the future? Do you have a sense of taking this powerful capability and pointing it in new directions, perhaps into supply chain, the ecosystem of partners, perhaps even into internal operations? What's the next step?
Yen: It's actually exciting times, because Vertica has proved itself so well. It's also very cost-effective. One of the projects that we're working on right now is that we have a relational database for our MRP system. It's more of an ODS reporting system. We’re actively converting the ODS system, which is actually a replicated database of the relational database, into a Vertica database. We're able to kind of replicate, mimic the native database replication scheme on the relational side, and use Vertica for it.
It's a use case that we were a little skeptical about in the beginning. Could this be done in Vertica? We thought the payoff would be great if we could do this on Vertica, the speed for performance, the storage footprint, would be amazing. So far, it's turned out very well for us. We’re still in the middle of it, but all things point to success there.