News Stay informed about the latest enterprise technology news and product updates.

Goodnight: SAS data analytics offers much more than BI from IBM, SAP

In this exclusive interview, SAS Institute CEO Dr. Jim Goodnight says his company’s data analytics expertise separates it from BI competitors SAP and IBM.

Since its founding in 1976, SAS Institute has always been known as a data mining and statistical analysis specialist....

But for the last several years, SAS has also been making waves in the broader business intelligence (BI) market, scoring among the top BI vendors in all the analyst rankings.

Still, the company has sometimes struggled to get that message out. Meanwhile, BI heavyweights SAP BusinessObjects and IBM Cognos have embraced the idea of BI for the masses – extending relatively easy-to-use reporting and query tools to non-statisticians.

But the pendulum may be starting to swing SAS’s way again. As its competitors scramble to meet customer demand and add more sophisticated data analytics capabilities to their BI suites, SAS, thanks to its years in the analytics business, is already there, according to Dr. Jim Goodnight.

“I think that’s the big differentiator for us,” Goodnight, a co-founder and CEO of Cary, N.C.-based SAS, said in an exclusive interview with

Following is an edited version of the interview, conducted last week at the SAS Global Forum in Seattle. In it, Dr. Goodnight also explains his decision to enter the social media analytics field, how SAS determines which data warehouse vendors to partner with on in-database analytics, and why industry-specific analytic products are the right approach. SAS Institute’s reputation is that of an expert in deep, complex statistical and data analysis. But you also play in the more mainstream BI space. What kind of company do you view SAS as being?

Dr. Jim Goodnight: Well, we announced our own business intelligence product six or seven years ago, and by business intelligence I mean a simple reporting and query tool that is easy for the masses to use. That’s the one thing that we didn’t have prior to about six years ago. And that’s why we were always known as just a statistics and heavy-duty analytics vendor, because we didn’t have that very simple-to-use, low-end interface BI tool. We have that now, and we’re constantly working to improve it, to make the user interface better. We recently made a move to put a lot of our results into flash, a much more attractive output.

But beyond business intelligence, we want to be known as a business analytics company. You’re seeing Business Objects trying to get into that space. That’s why IBM bought SPSS, to try to get into that space. Business analytics is the hottest thing going right now. More and more companies are interested in business analytics. And the fact that we’ve been there for the last 35 years is of key importance. We have more in-depth analytic capabilities than any other company in the world, and we continue to expand. So when you’re pursuing a new customer, is that how you try to differentiate yourself from your BI competitors?

Goodnight: Yes. It’s the credible depth of analytics we have versus just basic business intelligence, which is pretty much all the other vendors have. Business intelligence is sort of a marketing term that was made up to say, well, we can’t call ourselves a query and reporting company, we need to call ourselves something different. Let’s call ourselves “business” … um … make it “business intelligence.” And suddenly people that do nothing more than look in the rear-view mirror and report on what happened yesterday are called business intelligence companies.

A lot of customers that we’ve talked to have been very disappointed in their BI vendor. They thought they were going to get a lot more out of their BI tool. That’s because a lot of people think of business intelligence as data mining. In fact, if you ask somebody what business intelligence is, I saw a survey once and the No. 1 response was data mining. Well, the BI tools are query and reporting tools. They don’t do data mining. I think that’s the big differentiator for us, plus our industry-specific solutions. Why do you think taking an industry-specific approach to business analytics is the way to go?

Goodnight: Well, take the financial sector, for example. The financial sector represented I think 42% of all of our revenue last year, so shouldn’t we really try to get in there and understand the problems facing that industry and then try to build solutions that are specifically tailored to just that one industry? And the same thing goes with pharmaceuticals and retail.

I think while Cognos and Business Objects just stuck to the one thing that they did – query and reporting – another differentiator for us is that we’ve moved on and conquered the industry-specific solution space. SAS recently announced a new social media analytics application. One of the knocks on sentiment-analysis products is that the underlying text-analytics software is not accurate enough. How have you overcome the accuracy problem to social media analytics, and do you think customers are ready to embrace it?

Goodnight: We’ve been working on sentiment analysis for the last two years to get that right. That’s the work that our text analytics group up in Boston, a company previously called Teragram, has been doing. I think a lot of our customers will want to try it and can do so without a huge investment. We’ll just host these applications ourselves. If they want to try it for six months and they don’t like it, they can just drop it. I think it’s a good way for people to dip their toes into the water and decide whether they want to go all the way in or not. SAS has also made a lot of news lately regarding its in-database analytics partnerships with data warehouse vendors, including Teradata, Netezza and IBM. How is in-database analytics helpful to customers, and how does SAS go about deciding which database vendors to partner with?

Goodnight: The shared-nothing databases – Teradata, Netezza, (HP) NeoView – all of these are candidates to move a lot of our data collectors down to the node level. Usually, in any of the statistics that we do, the very first pass of going through the data is to build up matrixes and things that we use and then do the forecasting with. Rather than pipe every single record up one at a time to a central processor to do these computations, that first phase, which is quite often more than half the job, can be done on each individual node sitting out there. We can then aggregate the results at the central node.

I’ll just give you an example. Let’s say you’ve got a total of nine variables in a regression analysis, and each node contains a million records. And let’s say you’ve got 96 nodes, each with a million records on it. To do regression analysis we only need to compute 45 numbers, 45 statistics, from all of that data. So it’s much easier for us to ask node 1 to send us those 45 numbers versus a million records. That’s sort of the concept of in-database analytics.

As for the data warehouse vendors we partner with, they’ve got to have enough buy-in to make it worth the cost because this is expensive for us to do. Also, there are some vendors whose architecture just is not very receptive to a foreign object down there in their database. A lot of these databases are tuned to the point that they can’t bear any extra cycles. It’s a little bit of a give and take, and we will partner with only those vendors that are truly interested in working closely with us and are willing to invest some R&D on their side to open up that node for us to force all of that stuff in. That’s the requirement. While here at the SAS Global Conference, I’ve heard from a couple of SAS customers that while they are happy with their decision to invest in SAS, they’ve encountered some performance issues that they wish SAS could have caught before deployment. What is your approach to this issue?

Goodnight: I think, historically, performance issues, more than anything else, are due to a lack of tuning in the particular environment at a customer site. They’ll be running an operating system with some option turned on or off which just messes up performance really badly. And there are a lot of those little switches that you can throw on in these environments that can cause that to happen. But isn’t it SAS’s job to address those issues?

Goodnight: Well, we always do. If it’s our software and a customer is having problems with it, we’ll help them solve it. Finally, in terms of pricing, how do you think SAS stacks up among its competitors in terms of price for value received?

Goodnight: I think our prices are just about right. We have more and more of the analysts saying our pricing is where it should be. I guess the thing that bothers me sometimes is that people have SAS in-house and they don’t fully make use of it. And I think that’s one of the things that people need to look at more closely. You know, why aren’t we using SAS more? It’s already here, we don’t have to pay any more for it, and we can use it all we want to at no additional cost.

For example, we have a number of mainframe sites that only use SAS to run MXG for capacity planning, when there’s so much more they could be doing. I think it’s mostly an awareness issue. So we created a group that went into effect on January 1 that’s going to be doing more communications with existing customers and talking about how they can improve their utilization of SAS, new things they can do, new ideas, new usages.

Dig Deeper on Business intelligence data mining

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.