SAN FRANCISCO — Within the confined EDA industry, startup companies have to choose one of two strategies: direct assault or a more patient, laid-back approach. Executives who found success with both philosophies came to exchange stories at the EDA startup panel during last week's 35th Design Automation Conference.

Dubbed "The EDA Startup Experience: The First Product," the panel focused on the merits and drawbacks of different product strategies for a newborn design-automation company.

New companies have to make their mark through technology, not just by selling cheaper software, most panelists agreed. But they differed on how ambitious a company's first product needs to be. Some, such as Curt Widdoes of 0-in Design Automation Inc. (San Jose, Calif.), suggested that startups have to go for the big score by developing a "franchise" product, something along the lines of Synopsys Inc.'s Design Compiler.

Others argued for a more gradual approach: picking a problem to solve, then allowing the product to develop into a franchise over time.

"You can start in a space that is not necessarily key or crucial, and over time it will grow," said Dean Drako, chief executive of Design Acceleration Inc. (San Jose). DAI's Signals can was an example, he said: It solved the simple problem of low simulation productivity and, as the industry grew, its importance grew as well. Eventually, the product became a franchise.

Still, considering the increasing dominance of EDA's largest vendors, some executives were convinced that a startup has to tackle big targets. Rajeev Madhavan, in particular, used that approach in building a business model for Ambit Design Systems Inc. (Santa Clara, Calif.). Rather than try to nestle into a corner of the synthesis market, Ambit launched a full attack on market leader Synopsys.

"I don't think you have a choice. Hit them head on," Madhavan said. "I saw a lot of synthesis startups who said, 'I'm cheaper.' I think we're in a market where cheaper doesn't matter."

Madhavan did not say whether he was following a similar route with his new company, Magma Design Automation Inc. (Palo Alto, Calif.).

Others disagreed with Ambit's approach, saying the bigger EDA vendors are too well-fortified. "You're foolish to start out your company with a charter of competing," 0-in's Widdoes said. Companies that do so successfully are exceptions, he said. "There's lots of customer needs. You want to go find the need that has the least competition."

Likewise, synthesis vendor and Ambit competitor Synplicity Inc. "chose not to go head-on with Synopsys," said Alisa Yaffa, founder and chief financial officer. "A niche is good if it's a growing niche. You can have a lot of business" without charging at the market leader, she said.

Unglamorous targets can also be lucrative for a startup that knows its market and its purpose. That was the case with Chronologic Simulation (Los Gatos, Calif.), which drew early scorn for developing a Verilog simulator.

"There was no way we would have gotten venture funding for Chronologic," cofounder John Sanguinetti said. "Of course, we were Verilog users. We knew people weren't going to switch to VHDL, at least not voluntarily."

Despite having never run a software company, Chronologic's founders were able to use their firsthand knowledge of the tools to create something users wanted. "Being able to make good engineering judgments is one of the things that set us apart from the other people making Verilog clones at the time," Sanguinetti said. "We knew what we could leave out."

Whatever the product's target, high margins are vital, Widdoes asserted, because they generate the cash to create a distribution channel. "The market is too small to allow a company to build a strong channel with low margins," he said.

It's especially important to forge relationships within the small EDA customer base, executives said.

"The EDA industry is nice because it's small enough that you can go talk to all the customers. There's really only 20 or 30 large customers, and you'd be foolish if you didn't go talk to all of them," Widdoes said. "We [at 0-in] actually altered our product during that procedure."

The panelists did agree that the product has to fit into the existing design flow, at least in the beginning. Few companies will be willing to sacrifice existing libraries and scripts for a startup's technology, they said.

It's also important to chase down an existing need or deficiency, because pure technology won't find buyers in the EDA market. "A lot of people are trying to create a market, but that's a tough road to follow," Yaffa said. High barriers to entry are also a must, she said, to prevent the larger EDA players from intervening with their own products.

"A lot of times, you'll see an EDA company or a startup company with some interesting technology looking for a problem to solve," Drako said. "That's not the right approach. Pick a problem that people are really facing," one that's consuming engineers' time.

Making the shift from engineering to sales is also a key, panelists said, considering most EDA entrepreneurs come from the technical side of the field. One trick Synplicity discovered was to give away its software, which was small, to field engineers to use on laptops to demonstrate design flows. That created a "virtual sales force" that helped get word out, Yaffa said.

"As entrepreneurs, we are salespeople. The first customers will expect contact with the entrepreneurial team," said Lorne Cooper, chief executive officer of Sente Inc. "Selling to them is a very personal experience. You've got to get out there, reach out and touch the customer."

In addition, panelists unanimously stressed the importance of building the core startup team. Personalities do matter, they said, and can determine life or death for a company when hard times hit. "If you have any issues in the top management or the core team, you're doomed from the start," Drako said.

"All startup companies, including EDA startups, go through problems. A great team will eventually work its way through the problems," Widdoes said.