Well, SAP says they're going to leave BO [Business Objects] alone and let it run on its own, so I don't know that there's a whole lot of change. I think the major change is that SAP finally acknowledged that its Netweaver BI strategy was a failure so they needed to go out and buy something.
There was a lot of talk in that deal about not the "power to know" but "the power to do." Do you have any reflections on the move toward embedding analytics, prediction, BI, rules, and so on directly within applications?
Typically in the BI space in an operational system, we would use a decisioning engine in which we have embedded rules that we've computed, generally offline over some period of time so we could build up models. That's precisely what we're doing at HSBC. It's an operational system that evaluates whether a credit card presented to a merchant is fraudulent. We do it through massive amounts of analysis on the historical data associated with each individual [account]. We have about eight different categories of users, and for each one we build a neural network model involving about 500 variables. When the card is presented, we're sent the information and we score the transaction. Most of the real-time BI or real-time analytics that you hear about is really just scoring, which is evaluating a model that has been previously computed.
You recently announced a deal with Teradata to embed analytics into its database, and Netezza recently announced that SAS would be among several partners embedding analytics within its appliance. Do you see this spreading to other high-end data warehouse environments?
The beauty of Teradata or Netezza is that they have shared-nothing environments, which lets us deploy our analytics into each compute node. Instead of bringing all the data into one compute node to do the computing, we can now do it at 100 nodes, if that's how many nodes you have in the grid… We're going to invest heavily in R&D with Teradata to move our applications down to that level. There are a lot of them, like our anti-money-laundering application, that will fit very nicely in a Teradata environment. Each node will own a particular set of accounts, and all the computation for anti-money-laundering can be done on that particular node. If you have 100 different nodes, we can run the jobs 100 times faster than we're currently running them.
Two years ago at the SUGI (user group) conference in Philadelphia, you put a big push on query and reporting, a part of the BI market that SAS and Cognos now seem to be downplaying. Has the market changed that quickly?
We began working on our easy-to-use query and reporting tool about seven years ago. We released it three years ago, and since then we've risen to number two in the BI market right behind BO. Cognos and Business Objects particularly had been touting their query and reporting tools as "business intelligence." That was somewhat confusing to the market because people normally think of BI as things like data mining, forecasting and true intelligence...
So, we weren't recognized as a BI vendor three years ago, but now we're recognized as number two. Then, all of a sudden, everybody says, all the BI vendors are going to be consolidated, what's going to happen to you? There's the frustrating part because people don't understand that when we talk about BI, we're talking about the incredible depth of analytics that we have.
The HSBC credit card fraud deployment is just one example. If you go to another BI vendor as ask them for the probability that a customer will leave you next month, they can't do it because they can't drill down and find that. It's not in the database; it has to be computed. We're number one in anti-money-laundering. I consider that BI. We're number one for credit risk for Basel II and we've been ranked number one for operational risk. That's all built on heavy-duty analytics.
There's a lot of talk about performance management these days. How do you define it and will SAS put more of an emphasis on performance management?
We've had a performance management product for years — ever since Kaplan and Norton came out with their balanced scorecard methodology. The question is, are people using the methodology or are they just using the dashboard as an EIS? We used to call performance management "executive information systems," but that term has just completely disappeared. Dashboards are no more than a display system for some of the KPIs and metrics that companies are looking at. Is performance management just a scoreboard display? If so, then it doesn't amount to much. Anybody can do that.
So what has SAS put the focus on in its product, and will performance management become a more important area?
We believe in the Kaplan-and-Norton methodology where you're not just looking at finance, you're looking at customers, you're looking at employee satisfaction — everything. As I say, we've had the product for about ten years and it does fine.
We're much more into the high-end analytics that nobody else does. We have customers like Kohls, which has something like 1,500 department stores in the US. They're growing rapidly, and they're extremely dependent upon us for sales price optimization. Some of the financial analyst reports on the company actually talk about the use of our software. Optimization involves linear and nonlinear programming problems where you'll end up with as many as two million rows and five or six million columns in the optimization. No other vendor is doing that.
You've been asked before and with every vendor acquisition you're asked again, but what's future of SAS as a private company and is there a succession plan in place?
Succession plan? We've done that already. The board of directors meets regularly and discusses that. So far they're very pleased with my performance and want me to stay around. [Goodnight winks]
And SAS's future? The standard line from your marketing people is always "we can't serve our customers any better as anything other than an independent company."
Sure, I guess that sounds good. Our customers seem to be quite happy with us.