Swiffering new EDA tools

Why isn’t a new EDA tool like Swiffer?

One point that I’ve made before is that big EDA companies suffer from being unable to get new products into their channel. As I said earlier:

“When so much is riding on just keeping the customer on-board using the existing tools, the salesperson becomes very risk averse about selling new products.”

The effect of this is that big EDA companies can only sell to customers once there is market demand.  But that is the same problem as Proctor and Gamble faced with, say, Swiffer. Nobody was demanding mop with replaceable sheets, nobody knew one was available. So traditional marketing showed how useful it could be and that it was available at your local supermarket and now Swiffer is on track to be a billion dollar business.

Why can’t marketing do much to create demand in EDA?  I don’t entirely know, but here are some plausible relevant things.

Firstly, the EDA market (for IC design, not for FPGA or embedded software) is inelastic. No matter how much advertising is done, no matter how low the price, no matter how appealing the packaging, the market for EDA tools is fixed. Sure, we can steal market share from each other, maybe we can increase ASPs, we can expand the definition of EDA. But there is no untapped market of people out there who never knew they wanted to design a chip, in the same way as we all turned out to be a market of people who never knew we needed a post-it note. So we are only marketing to people who already know they are designers.

EDA is not even like other software industries. It values different things because it moves so fast. All users complain, with justification, about the bugginess of EDA software, but they can’t get by with the old solid version in the same way as in slower moving software industries. In books like Crossing the Chasm and The Innovator’s Dilemma, marketers are told to worry about the job that the customer hires you to do. The customer doesn’t want a drill, they want a hole. The job is holes. But as Tom Kozas and Mike Sanie point out here, when the EDA engineer goes to Home Depot, he’s not looking for ways to make a hole. He’s already decided that he wants an 18V cordless drill with two gear ratios. Maybe he’ll pick between DeWalt and Bosch but he’s not looking at those ads for explosive nail-guns.

Next, the design engineer has been burned before. Because the technology treadmill moves so fast, tools don’t always work well (or sometimes at all) but the purchaser doesn’t have the luxury of waiting for code to mature, for standards to be in place, for the landscape of winners and losers to be clear. But a lot of IC design is about reducing risk (because we can’t just fab the chip repeatedly in the equivalent way to a software engineer compiling and testing the code). One component of risk is using a new tool so there is always a push of potential advantage of the new tool against the pull of potential disaster if it fails. So designers have learned to evaluate new tools in enormous detail, to understand not just what they should do, but what they actually do and how they work internally to do it. Other people don’t take the cylinder-head off the engine before buying a car.

Brand name counts for very little in EDA. To the extent it counts for anything in this context, it stands for a large organization of application engineers who can potentially help adoption. It certainly doesn’t stand for rock-solid reliability. The speed of development means that every large EDA company has had its share of disastrous releases that didn’t work and products that never made it to market. There are no Toyotas and Hondas in EDA with a reputation for unmatched quality. I don’t think anyone knows how it would be possible to create one without it also having a reputation for the unmatched irrelevance of many of its products due to lateness.

So there are a few theories. Like all stories after the fact, they are plausible but it is not clear if they are the real reason. But the facts are clear: traditional marketing, such as advertising, doesn’t work for EDA products.

This entry was posted in eda industry, management. Bookmark the permalink.

Comments are closed.