SAS Chief: Hot on Fraud Detection, Cool on Cloud Computing

At a recent media event, SAS CEO Dr. Jim Goodnight spoke with IDG Enterprise Chief Content Officer John Gallant about how businesses did -- and didn't -- take advantage of BI during the downturn and how the economy has changed the playing field in the BI market.


Was there ever a year when corporate America -- particularly Wall Street -- needed more insight and intelligence about how to navigate stormy economic waters than 2009? Dr. Jim Goodnight co-founded (along with John Sall and other North Carolina State University colleagues) SAS in 1976 and has led the $2.8B company to a leadership position in the business intelligence (BI) and analytics market. At a recent media event, Goodnight spoke with IDG Enterprise Chief Content Officer John Gallant about how businesses did -- and didn't -- take advantage of BI during the downturn and how the economy has changed the playing field in the BI market. Goodnight also talked about why he's not overly enthusiastic about the potential for BI in the cloud computing environment and why the perception lingers that BI has failed to deliver for IT and business.

Business Intelligence Definition and Solutions

The financial industry is a big market segment for SAS. What lessons do you think were learned in the financial industry as a result of the downturn?

I think beyond any doubt that the people who were doing risk computations or relying on the ratings provided by Moody's and other folks on some of this junk stuff that they were calling Triple A never really got down inside of those vehicles to see exactly what was in there. They just lumped them all as part of the Triple A paper. And it wasn't until a few of them started going bad that they began to realize, well, our risk is a lot bigger than we thought it was because we have been misclassifying the risk on these particular types of paper. I would have loved to see them get out of some of the derivative stuff that they're doing -- especially these exotic derivatives that nobody quite understands exactly how they work. I would love to see banks back into the traditional role of being a depository for the nation's families.

So the information was there but people were making the wrong decisions?

The information was there -- they just didn't dig deep enough. There was a recent article about how Wall Street lied to their computers. I didn't read the article. I didn't need to. As soon as I saw the headline I said: 'I know exactly what that's about.' That was the biggest problem we had with Wall Street; they were using the wrong ratings on all of the instruments they had.

How has the past year changed the market for SAS? What kinds of new opportunities have opened up and what have you seen in different market segments?

We're seeing a lot of interest this year in fraud detection. We've got probably 10 applications that we're already working on and another 10 or 20 in the pipeline. In times like this banks and insurance companies have got to try to make sure every dime we spend is properly spent. In a down economy when people really need the money, more and more people turn to fraud. That's been the biggest thing we've seen and that's across a lot of industries: insurance, healthcare, welfare fraud, tax fraud. It's just a huge, growing area for us.

You mentioned in a presentation two segments that are growing at a nice clip -- retail and government. What kinds of applications are you seeing in those markets?

In the government, we're seeing more and more use of technology, especially in the agencies that are going after terrorists and trying to find the bad guys, as well as for logistics in the military. In retail, it's the whole optimization process -- site optimization, pack optimization. Many chains order the same pack of, let's say, sweaters and every store gets the same thing. We've been able to point out to them in so many cases that there are a lot of variations in the population where the stores are located. This particular store might have a history of running out of smalls and this one over here may have a problem of always too many smalls at the end of the season and they go on clearance. If you can balance that and get just the right sizes going to each store the customer is going to be happier, because when they go in there they're more likely going to find the right size, and the merchant is going to be happy because at the end of the year, he's not having to put all this stuff on sale that was mis-sized really for that particular store.

[Chief Marketing Officer] Jim Davis made a comment during a presentation that there are too many companies in the business intelligence market. So, how do you differentiate yourself from the pack?

It's our depth of analytics and our deep involvement in the banking space, from credit card fraud to risk computations -- credit risks, market risks, operational risks. We continue to pour a lot of money into risk. I've been working on some risk computational projects myself to try to speed up risk computations, especially when it comes to stress testing.

When you think about some of the bigger competitors -- the IBMs, the Oracles, the SAPs -- how has the market changed in the past year? How has the competitive dynamic changed because of the economy?

The biggest change is that we're not seeing Cognos as much as we used to. We're seeing a little bit more competition from Oracle now in the pharma space.

SAS CEO unfazed by IBM analytics challenge

They're really pushing life sciences in some of their acquisitions. But, remember, we compete with 200 or more different companies in the anti-money laundering, in the fraud stuff and the risk stuff. We won't see any of these [players] in that space. Just because some big companies bought some of our smaller competitors hasn't really changed the landscape for us that much.

Speaking of buyouts, your company acquired IDeaS, which provides revenue management via the software-as-a-service approach. Can you describe SAS's overall cloud strategy?

First of all, I don't see anything greatly new and wonderful and different about cloud computing. It was timesharing way back in '60. It's not a whole lot different. I certainly have issues asking a bank to send us all their data and we're going to put it up on a cloud. They're going to say, 'What about security? How will I know who else is up there in that cloud?' I don't know, it's just a cloud. I think that from a security standpoint the use of clouds is a few years away before they can have the kind of security that people feel good about.

Do you think that the processing power that is becoming available in the cloud could open up analytics to markets -- like the SMB market -- where today it's not widely adopted?

We can provide that service right out of here. We've got plenty of hardware here. I think a lot of companies may be deciding whether they should be adding more servers themselves or experimenting with using clouds to see if that could curtail the cost of ownership of all the hardware. That to me is still a big unknown.

One of the knocks on analytics or business intelligence is the data integration problem -- the challenges that people have getting the one version of the truth among different data sources and types within the company and how that continues to limit BI. How real is that problem today?

We have our own data integration tools from our DataFlux company. We're on par with Informatica when it comes to data integration and data quality. But there is a problem with integrating data. So many organizations have data in so many different silos that it's really hard to get at all of it. Just the sheer process of merging different databases together that have used different rules for how to put down a job description or a product -- whatever -- just trying to get all that merged together is a large [challenge] of BI.

Is that what leads to the sense that BI disappoints? We see it, for example, on a list of CIO top priorities year after year, meaning there's some sense that it hasn't delivered on the promise yet.

I can understand that. About 10 years ago, we decided we should spend some money and get into the BI business. Before that, we had always been predominately analytics. People have used SAS for reporting for the last 35 years, but the fact is that really easy to use interfaces for the department level, we didn't have it, where Cognos and BI did. So, we decided that we ought to go ahead and develop our own and that's why we've been in the business now for the last six years because we spent four or five years developing stuff. So, we're somewhat new to what is the so-called BI market. Now, BI as it grew up was really nothing much more than creating reporting tools. You know, we help you formulate a query doing some dragging and dropping and then the BI vendor creates statements and submits them to the database to run and then produces results and puts them on the screen.

BI was really never ever anything more than just creating reporting tools. So, obviously, if you're expecting intelligence to come out of a BI tool, you're a little disappointed when you discover that all it really does is reporting -- no heavy-duty analytics, no forecasting the future, no helping you optimize your processes. Where SAS differs is that we've had that part all along and then we decided to push down into that lower space of creative reporting. Now we're thought of as one of the big BI vendors and what are you going to do? People are disappointed in BI -- well, of course they are -- there's no analytics involved.

Do you think customers understand that differentiation between business analytics and business intelligence?

No. I think most people think that business intelligence means data mining and stuff like that.

What's the state of real-time analytics today? The vision that analytics can drive business processes in real time.

What most people are talking about is the fact that they are executing models in operational systems to help make the decisions about what to do next and where to go. It's not rocket science to do real-time analytics. All you're doing is evaluating a model to produce a score. Now, you've done the work on those models offline and spent months of effort in developing the most precise model you can to predict, say, someone's [credit] score. Then you're just applying that model in your operational system. We've been there for a number of years. Every credit card fraud system uses models to evaluate and score. FICO scores are computed like that. There are an awful lot of call centers that are using real-time decision management to make decisions.

Where do you see the market going in the next five years? What are the key things that you would like to get the company to do and deliver for customers?

There's no question that we're going to be dealing with larger and larger amounts of data and larger and larger problems that have to be solved. Some of our operations research that used to be 100,000 by 200,000 -- you know, 100,000 rows by 200,000 columns -- now we've got 2 million and 3 million. Things have grown almost a thousand-fold over the previous sizes. Well, computers are not going to get a whole lot faster in the near future until we get off of silicon and ultraviolet wavelengths of light that determine the size of the etching. So, they're adding cores -- putting multiple cores on a single chip. We're going to take advantage of that for some of the high-performance work. We have to learn to take advantage of the fact that we can't rely on a single CPU -- we've got to always be thinking in terms of doing things in parallel and using multiple CPUs. It's a totally different way of thinking. All of us growing up in computing learned how to do things sequentially. You finish this block of code and then you go to the next block of code and it runs and it's just one block of code after the other. Now, you have to think in terms of how to use multicore processors if you're going to tackle the really big problems.

And does that involve a change to the entire product portfolio to take advantage of that?

No, you're going to choose the processes that need to be speeded up. How do I speed up those processes in a grid environment where you're using all the machines simultaneously? There are not that many problems -- certainly opening up your e-mail doesn't need it.

Speaking of e-mail, how do we deal with the mass of unstructured data that grows every day? How do we take advantage of that?

Well, we use Teragram, our natural language processing solution. We're working with some of the banks and trying to analyze all the data that comes into their call centers. Let's say the bank has people with problem loans that they're talking to. We can go through and categorize let's say e-mails or call center data based on how the person is coming across. Are they mad? Are they happy? Are they pleased? That's finding a lot of uses.

You've been doing this a long time. What did you learn in the past year? What did this particular economic situation teach you?

I think it's really how appreciative people are that you're able to tell them that there will not be any layoffs -- and stick by that. People are more productive because they see what's happening out there on the street. Some of their friends have lost their jobs and are losing their houses. I made a pact with [employees] in the beginning of the year that we would not have any raises this year, but we will also not have any layoffs.  Everybody was very pleased with that concept.

This story, "SAS Chief: Hot on Fraud Detection, Cool on Cloud Computing" was originally published by Network World.

Copyright © 2009 IDG Communications, Inc.

7 secrets of successful remote IT teams