The MathWorks: Adding value

The MathWorks has built its reputation on elegant tools. Can the company take itself to the next level?

by Erik Sherman -- Movers and Shakers, 8/15/2003

THERE ARE ENGINEERING firms that add sophisticated marketing and eventually became household brands. And then there are engineering companies that remain engineering companies, but have such strong products that they become household names--or at least "officehold" names. The MathWorks is one of the latter.

The brainchild of both CEO Jack Little and chief scientist Cleve Moler, The MathWorks is largely what its name would suggest: a place where math comes first. But this isn't about exotic calculations that interest only a handful of theoreticians. Instead, the company's major products, MatLab and Simulink, are popular and versatile research and engineering tools  that perform a range of functions, from iterative design of complex systems to testing hardware and generating solid code.

Jack Little, CEO, The MathWorks

That's quite the technical hat trick, and there's a reason that The MathWorks is the leader in its particular market space. But the company also faces challenges: limited resources, potential budget constraints among customers, and a failed acquisition that could, under the right circumstances, turn into a significant competitor.

It can be difficult to get information about the company, which is not only privately held but also largely shuns the spotlight. Little almost never gives interviews. As he explained in an address at the Massachusetts Software Council, "Most of what you read about in the newspapers is about companies on the IPO track....It took me a long time to figure out that, in fact, the IPO track is only the most visible game, and I shouldn't be playing that game if that's not what my goals are."

So The MathWorks remains invisible, except to the legions of engineers who first used the software in school and then requested copies on the job. "Most of the growth is coming from the engineering graduates, so you're assured of a growth market pretty much every year," says Daya Nadamuni, principle analyst, design and engineering software, Gartner Dataquest. "Even if our engineering graduates drop, there are other parts of the world that will graduate engineers."

"You can't fault success," says Dr Jerry Krasner, vice president and principal analyst with American Technology International. "I personally think that they could be doing much more business. Jack runs the place as an engineering shop, and they're brilliant. I'd run it as a marketing shop. But what the hell do I know? I'm not as rich as him."

The MathWorks took in more than $220 million last year, according to The MathWorks fellow and chief strategist Jim Tung. The company has 1,000 employees and roughly 600,000 users, and has been profitable "since day one." While the numbers are respectable, they hardly imply resources that could finance a rapid expansion. But that doesn't interest management anyway. "Growth in itself is not a goal for us," Tung says. "We've grown in a carefully controlled manner."

Although early exposure in engineering schools is a powerful sales tool, Krasner argues that the real clincher is economics. His company regularly surveys embedded developers and asks how close final designs are to pre-design expectations in functions, performance, features, and schedules. In the last two years, a little over a quarter were not within 50 percent. As of last March, that number jumped to a third. "Designs are getting much more complex...and the final result gets within 20 percent of expectations only 30 to 40 percent of the time," Krasner says. "This is astounding. Think of the cost."

That cost creates a powerful incentive to find tools that make the design process more predictable. And that, ultimately, is where The MathWorks does its banking. The company has some competitors, but as Nadamuni puts it, "They don't have the breadth that The MathWorks does. You can't say that there's an Oracle and SAP going head-to-head because they can offer similar functionality."

Given the circumstances, marketing may be redundant. "I'm not sure that it even matters, because they are so well known," says Chris Lanfear, embedded software group manager at analyst firm Venture Development Corp. However, should the company decide that significant growth is desirable, or even necessary, the lack of marketing may cause a problem. "To take the next step to be a larger company, seen as a strategic partner to their customers, I think you have to be a little more out front," Lanfear says.

Would the market support a bigger The MathWorks? It might, if the company offered more specialized tools--variations on the two flagship products--that could satisfy additional user needs. In some of its core markets, the company does offer "toolkits," or add-on packages with specific capabilities. For those not in the core markets, The MathWorks relies on third parties that can develop specialized toolkits. "We have many opportunities we could pursue," Tung says. "Our approach is how to leverage the vendor community."

One problem with the "engineering mentality" is that product capabilities and quality are not necessarily enough to win over users. Customers have been slower than the company would like to adopt some of its approaches. "In many areas...[customers are] used to rolling their own," Tung says. "When I think of growth of markets, that remains the limiting factor."

Code generation is a fine example. "The issue is to convince the OEMs that it's production quality," Lanfear says. "It's a matter of changing some mindsets in the markets they sell into." And that is a job for--surprise--marketing.

So the company might have to come to terms with the need for not only marketing, but also public relations. Consider the company's attempt to buy MATRIXx from Wind River Systems. The product competes with some MathWorks offerings, but was "a relatively small product and diversion" for Wind River, according to Lanfear.

Wind River struck a deal to sell MATRIXx to The MathWorks, which would maintain it but not do new work on it. "That was fine, except they were essentially giving it to The MathWorks to kill it," Lanfear says. MATRIXx had many dedicated fans in the military--a key MathWorks market--who became angry and complained to the Department of Justice. And so, in June of 2002, The MathWorks found itself as one of two defendants in a lawsuit alleging violations of the Sherman antitrust act.

The action was eventually settled when National Instruments bought the product. This well-financed company--more than 50 percent larger than MathWorks and with a public status that allows it easier access to capital--has its own significant technical resources and has decided to move into the modeling business. The military support for MATRIXx gives National Instruments a good base from which to build.

The MathWorks still has a substantial head start, but the company may find that to continue its dominance, it might need the glad hand as well as the drawing board.