Startups grow up
The first customer on a startup's dance card is critical
by Bill Roberts -- EDN, April 1, 2006
|
Sections:
The customer conundrum The case for a big customer The case for a startup customer Take what you can get  Shiv Tasker, CEO of Bluspec, is sitting pretty in start-up city: His first two customers were a blue-chip company and a fabless startup. |
Shiv Tasker, the CEO of Bluespec, a Waltham, Mass.-based startup that develops system-level-design synthesis tools for ASIC and FPGA chips, might be in the best position an entrepreneur could be in less than three years after launching. Ten months ago, he signed his company's first two customers: One is a large, well-known global semiconductor maker; the other is also a startup that is betting its success on the Bluespec tool set.
By first signing up a division of STMicroelectronics, Tasker has a blue-chip company as a paying customer he can use to lure other big companies. In fact, since signing up STMicroelectronics, in June 2005, Tasker has added three other large customers he considers key to Bluespec's long-term success: Texas Instruments, Analog Devices and Nokia. For the most part, however, these large companies have used Bluespec tools only tentatively, taking a cautious approach to an untried technology.
That's why customer No. 2, Element CXI, a Milpitas, Calif. fabless startup in stealth mode, is so important. Element is using the Bluespec tools all the way through to the tape-out of its first chip. This will give Tasker a real product to point to when he makes sales pitches.
"An STMicro or a TI has lots of existing IP," Tasker says. "(Element) is going to start by using us to design new IP blocks, and we will always get questions from potential customers about how much of their work done with our tools went to tape-out on a chip. With Element, I'm probably going to lose money on the deal, especially on support, but when it finishes, I will be able to say, 'Here is a full chip done with our tools.'"
Hiring talent, developing technology and raising money are all crucial to the success of a startup—just ask any of the CEOs included in this year's ELECTRONIC BUSINESS Best Small Companies list (below).
One other point the CEOs will make is that no startup is a real company until it has customers.
"Getting that first customer is one of the most critical steps in a startup's life," says Sanjay Subhedar, managing partner at Storm Ventures, Menlo Park, Calif., and a former entrepreneur. "We call it a 'teaching customer.' When you are developing a product, you want at least one customer, if not several, helping you and telling you the problems they face. You want a mentor, just as you want a mentor in your career."
The customer conundrum
CEOs of startups, their venture partners and serial entrepreneurs agree that the first customer is crucial. There is not wide agreement, however, on whether the first customer must be a blue-chip company—an STMicroelectronics or a Nokia, for example—or whether another startup will do. In fact, there are arguments for each. At the end of the day, most entrepreneurs are disinclined to turn down any first customer—although there are times when even that might be the proper strategy (see "Advice to startups," below).
In many cases, entrepreneurs cannot choose which camp the first customer falls into. They take what they can get. But when they plan initial sales strategy, or if they are given a choice between a tier 1 company and a startup, entrepreneurs face a conundrum. On one hand, large, established companies are by nature risk-averse, slow to adopt new technologies and very demanding when they buy from a startup. Startups, on the other hand, are risky as customers; can suck up valuable resources in support; and if they do succeed in using the technology, may not have much credibility with other potential customers.
"An enterprise-level customer wants to be cautious," says Mark Dubovoy, a partner at Leapfrog Ventures, Menlo Park, Calif. "They worry about whether your little company will be in business tomorrow or six months from now. Will it get the right level of support? How solid is your product? You may also have to give it the stuff for free initially in exchange for using it as a reference."
The alternative of having a startup as your first customer generally doesn't do you much good, he adds.
"You don't gain much credibility with the market. You compound your risk dramatically—now you are a startup relying on a startup," he says.
Dubovoy usually urges his startups to find a large company as their first customer. "You are better off starting small with a brand-name customer, in one division or department, than a decent sale with a startup or even a medium-size company."
The case for a big customer
Two entrepreneurs who agree that a large company should be your first customer are Patrick Soheili, who recently became CEO at Cradle Technologies, Sunnyvale, Calif., and David Fritz, the San Jose-based CEO of Silistix, Manchester, England.
"You want to work with a tier 1 customer, but you are looking for a tier 1 customer whose vision matches yours. Its road map must match the attributes your product offers," Soheili says.
Founded in 1998, Cradle is a fabless company that develops multicore DSPs, including a 16-channel digital video recorder on a chip, for surveillance and other imaging applications. When Soheili joined as senior vice president of sales and business development, in 2003, he knew that Cradle's chips would have applications in several markets, including surveillance, high-end imaging and videoconferencing. His first task was to narrow the field to a single market that offered the biggest earliest payoff. He analyzed the competitive landscape and talked to two or three executives at tier 1 companies in each potential market.
It took Soheili about six months to determine that surveillance was the first market to attack and that South Korean telecom companies, which led the world in developing networked intelligent surveillance systems, were the key potential customers. It's no coincidence that Cradle's first customers were Korean Telecom and one of its Korea-based video camera suppliers, Digi-flower.
“You are looking for a tier 1 customer whose vision matches yours.”
—Patrick Soheili, Cradle Technologies
Soheili says it is crucial to understand the strategy of your target customer, whether it is a tier 1 or tier 2 company, but this is even more important if you want to attract a tier 1 customer. "You might influence a tier 2's strategy, but you can't influence a tier 1 company's strategy."
Silistix's Fritz agrees. He says that large customers are worth pursuing initially, because "they are like dominoes. You knock one down, and you use that to get two more, and that is the only way to build a customer base." But snaring the first customer from among tier 1 players is a huge challenge, and preparatory work is necessary, he says.
Founded in 2003, Silistix was spun out of the computer research labs at Manchester University, in England. The company develops design and synthesis tools that chip makers use to create efficient interconnects on increasingly complex SoCs. Silistix compares the way these tools allow data to be transmitted across complex computer chips to the way data is transferred among computers on a local-area network.
Silistix has one nonpaying customer, Fritz says. But it is an important relationship. "This is really a beta customer. Sometimes your best customers are nonpaying customers, if they are serviced well, are happy and become references for paying companies," says Fritz.
The customer—a major semiconductor maker that declines to be identified—has been important to the company's evolution and in time will help Fritz secure new paying customers, he is certain. "We don't have a No. 2 customer yet, but we are close. We're expecting several new customers in the next few months."
“I knew there was nothing that could match the proof of a real tape-out.”
—Rajeev Madhavan, Magma Design Automation
Fritz initially identified a dozen large semiconductor companies he thought could be early users of his company's technology. About one third were Japanese, and Fritz knew from experience that they were ultracautious and would take a long time to sell to. He did not ignore them but scratched them from his list of top prospects. Likewise, he knew that the European companies on the list would not be as quick to adopt the technology as North American companies. He immediately cut the list to six companies in North America. From the intelligence gathering he did through his personal network of entrepreneurs and others, he determined which of the six were least encumbered by bureaucratic procedures and came up with a short list of three. "We ranked those and then went after them with all the ammo we had."
Did it ever cross his mind to find a needy startup company first?
"I would not have been terribly interested in a startup customer first, because it would have taken too many resources," Fritz says.
The case for a startup customer
That may be true, but one entrepreneur's poison is another's protein. A startup that bet the company on his startup's technology proved to be the perfect customer for Rajeev Madhavan, the CEO of Magma Design Automation, Santa Clara, Calif.
Magma, which develops EDA software, was founded in 1997, went public in 2001 and has challenged established EDA companies ever since. In its most recent quarter, which ended January 1, 2006, Magma reported revenues of $41.3 million. In four of the past five years, it was on EB's list of best small companies, based on revenue growth rates.
Magma's first customer was 3Dlabs, which was developing high-end graphics processing chips and is now a wholly owned subsidiary of Creative Labs, Milpitas, Calif. 3Dlabs turned to Magma only after going to its larger competitors, which referred it to the startup. "3Dlabs had already missed on a bunch of designs," Madhavan recalls. "Either it was going to do it with us or go out of business."
Magma had been making noise in the market, but its larger competitors had chided it for having no tape-outs to point to. Magma needed to prove to the chip-making community that it could deliver a full design.
"We stopped trying to sell to others for a few months and put our entire focus on 3Dlabs," Madhavan says. "Our R&D people became its designers. It was our chip as much as it was (3Dlabs') chip."
Madhavan had to defend his decision to board members and some of his own employees, but stuck to his guns. "I knew there was nothing that could match the proof of a real tape-out."
“We had several personal connections, which reduced the fear, uncertainty and doubt of having a startup sell to a startup.” —Chris Rowen, Tensilica
There was a downside, however. "Payment was conditional on getting the tape-out done. We lost momentum with a few other customers and had to walk away completely from another big potential customer. And my whole sales force quit on me because of this."
As a result of collaborating with 3Dlabs, however, Magma worked out the bugs in its software. Soon after the 3Dlabs tape-out, Magma's business was rolling. "Three months after we finished the 3-D chip, we had a spate of tape-outs. Among them was one by Infineon, our first big customer," Madhavan says.
Madhavan vows he would do it again. "Ours is not a common tale, but if I were to do another startup, I would look for a similar situation. Most of the board members respect me today for what we did."
Take what you can get
The reality is that many entrepreneurs don't have the luxury of being choosy about first customers. Momentum is everything at a startup. A big customer means big momentum, but a startup customer might be better than no customer at all. That was the attitude of Chris Rowen, CEO of Tensilica, Santa Clara, which has made the ELECTRONIC BUSINESS "Best Small Electronics Companies" list in the past.
Founded in 1997, Tensilica develops configurable processor cores and recently launched a family of standard processor cores that are likely to compete with those of ARM. Tensilica has more than 80 licensees, including brand-name companies such as Cisco Systems, Broadcom and Conexant. But it was not always that way. The first customer, signed in 1998, was Silicon Spice, a startup later acquired by Broadcom.
"We had several personal connections to Silicon Spice, especially to its technical team, which reduced the fear, uncertainty and doubt of having a startup sell to a startup," Rowen recalls. "It was also working on a fairly impressive product, a network access platform with a multiprocessor chip."
Rowen would have preferred a big customer first. "Anytime you sell to a startup, there is a theoretical risk. You won't know what it is like to sell to a big customer until you have a big customer. Their needs are different."
But Rowen has seen entrepreneurs spin the wheels of their own mind, awaiting the perfect customer. He thinks it is dangerous for a startup to become too enamored of its own thinking about its technology and business model and the problems it can solve. Better to work with a smaller customer, even a startup, to get out of the realm of theory and into the realm of practice.
"You may miss the boat a little bit by working with a small customer and not a big one," he reflects, "but working with a small customer is going to mislead you less than your own imagination."
Did your startup company have a blue chip company or a startup as its first customer? Send your story to feedback@eb-mag.com.
30 Best Small Electronics Companies
Ranked by annual revenue growth 2003-2005
| Revenue ($ millions) | Net income ($ millions) | Quarterly revenue ($ millions) | |||||||||||||
| Rank | Company | Description | FY 2005 | FY 2004 | FY 2003 | Annual revenue growth % | FY 2005 | FY 2004 | FY 2003 | Annual net income growth % | 12/31/05 | 9/30/05 | 6/30/05 | 3/31/05 | Web site |
| SOURCE: REED CORPORATE RESEARCH N/A = Not available/not applicable |
|||||||||||||||
| 1 | PortalPlayer | Semiconductor design | $225.2 | $92.6 | $20.9 | 228.3% | $48.2 | $10.4 | -$8.0 | 115.2% | $78.2 | $57.9 | $44.6 | $44.5 | www.portalplayer.com |
| 2 | Sonic Solutions | Multimedia, graphics, publishing software | $90.6 | $56.9 | $32.7 | 66.5% | $2.5 | $11.1 | $2.5 | 0.0% | $40.8 | $31.9 | $35.5 | $35.6 | www.sonic.com |
| 3 | Advanced Analogic Technologies | Semiconductors | $68.3 | $51.3 | $26.5 | 60.5% | $2.1 | $15.2 | $0.9 | 52.8% | $21.6 | $18.0 | $14.5 | $14.3 | www.analogictech.com |
| 4 | Tessera Technologies | Semiconductor packaging | $94.7 | $72.7 | $37.3 | 59.3% | $31.4 | $59.1 | $9.4 | 82.8% | $24.6 | $19.7 | $22.5 | $27.9 | www.tessera.com |
| 5 | FormFactor | Semiconductor testing equipment | $237.5 | $177.8 | $98.3 | 55.4% | $30.2 | $25.2 | $7.5 | 100.7% | $71.8 | $62.4 | $52.3 | $51.0 | www.formfactor.com |
| 6 | I.D. Systems | Wireless tracking systems | $19.0 | $13.7 | $8.0 | 54.1% | $0.9 | $0.4 | -$1.2 | 50.0% | $6.0 | $5.7 | $4.2 | $3.0 | www.id-systems.com |
| 7 | SiRF Technology Holdings | GPS design & software | $165.2 | $117.4 | $73.1 | 50.3% | $29.4 | $30.7 | $3.6 | 185.8% | $54.4 | $48.3 | $35.4 | $27.0 | www.sirf.com |
| 8 | CalAmp | Microwave amplification & conversion components | $220.0 | $128.6 | $100.0 | 48.3% | $8.1 | $5.7 | $5.2 | 24.8% | N/A | N/A | N/A | N/A | www.calamp.com |
| 9 | SERENA Software | Software | $208.1 | $105.6 | $95.8 | 47.4% | $9.5 | $21.4 | $23.2 | -36.0% | $70.1 | $64.9 | $59.4 | $61.3 | www.serena.com |
| 10 | Atheros Communications | Semiconductors | $183.5 | $169.6 | $87.4 | 44.9% | $16.7 | $10.8 | -$13.2 | 24.3% | $53.1 | $45.8 | $43.4 | $41.2 | www.atheros.com |
| 11 | Unica | CRM, marketing, sales software | $63.5 | $48.7 | $31.3 | 42.4% | $4.5 | $3.5 | $2.5 | 34.2% | $17.6 | $17.2 | $16.4 | $15.3 | www.unicacorp.com |
| 12 | Pinnacle Data Systems | Computer systems | $44.6 | $34.4 | $22.9 | 39.6% | $0.9 | $0.9 | $0.5 | 34.2% | $15.9 | $14.3 | $6.9 | $7.5 | www.pinnacle.com |
| 13 | EPIQ Systems | Financial services, legal, government software | $130.8 | $125.4 | $67.9 | 38.8% | $10.9 | $9.8 | $8.7 | 11.9% | $35.7 | $32.0 | $31.6 | $31.5 | www.epiqsystems.com |
| 14 | Hittite Microwave | Semiconductors | $80.7 | $61.7 | $42.0 | 38.6% | $21.1 | $13.4 | $7.2 | 71.2% | $22.7 | $21.2 | $18.9 | $17.9 | www.hittite.com |
| 15 | Ixia | Communications testing equipment | $159.3 | $117.0 | $83.5 | 38.1% | $33.7 | $18.9 | $8.7 | 96.8% | $37.5 | $42.1 | $41.3 | $38.4 | www.ixiacom.com |
| 16 | Integrated Silicon Solution | Memory chips | $181.4 | $181.0 | $97.7 | 36.3% | $37.9 | $3.5 | -$28.1 | 229.0% | $51.1 | $61.5 | $53.7 | $35.7 | www.issi.com |
| 17 | Rimage | CD/DVD duplication equipment | $95.4 | $70.8 | $53.8 | 33.2% | $11.4 | $9.1 | $7.7 | 21.7% | $24.3 | $28.0 | $22.3 | $20.9 | www.rimage.com |
| 18 | LaBarge | Electronic and interconnect components | $182.3 | $131.5 | $102.9 | 33.1% | $10.9 | $6.9 | $2.2 | 122.6% | $48.1 | $39.6 | $45.1 | $44.8 | www.labarge.com |
| 19 | FARO Technologies | Measurement software & equipment | $125.6 | $97.0 | $71.8 | 32.3% | $8.2 | $14.9 | $8.3 | -0.6% | $34.5 | $32.6 | $30.9 | $27.6 | www.faro.com |
| 20 | Cohu | Semiconductor testing equipment | $238.9 | $176.2 | $138.6 | 31.3% | $34.0 | $16.7 | $0.0 | 42.6% | $74.1 | $68.6 | $51.8 | $44.4 | www.cohu.com |
| 21 | Symyx Technologies | Materials research | $108.1 | $83.2 | $63.0 | 31.0% | $12.0 | $12.9 | $5.7 | 45.1% | $30.9 | $31.7 | $24.1 | $21.5 | www.symyx.com |
| 22 | Sirenza Microdevices | Semiconductors | $64.2 | $61.3 | $38.5 | 29.1% | $1.4 | $0.3 | -$6.2 | 116.0% | $19.5 | $17.2 | $15.3 | $12.2 | www.sirenza.com |
| 23 | Lowrance Electronics | Sonar & GPS devices | $146.4 | $111.9 | $88.3 | 28.8% | $9.4 | $8.8 | $4.6 | 43.0% | N/A | N/A | N/A | N/A | www.lowrance.com |
| 24 | ADE | Electronic test, measurement equipment | $116.9 | $88.6 | $71.2 | 28.1% | $40.9 | $8.8 | -$7.4 | 115.5% | N/A | N/A | N/A | N/A | www.ade.com |
| 25 | Semitool | Semiconductor equipment | $190.4 | $139.6 | $117.0 | 27.6% | $10.1 | $7.4 | -$21.2 | 16.8% | $55.3 | $47.8 | $47.4 | $46.3 | www.semitool.com |
| 26 | Advanced Photonix | Optoelectronic products & systems | $14.8 | $12.4 | $9.1 | 27.5% | $5.3 | $0.8 | -$0.8 | 157.3% | $6.5 | $5.2 | $5.1 | $4.0 | www.advancedphotonix.com |
| 27 | Macrovision | Content & digital rights protection | $203.2 | $182.1 | $128.3 | 25.8% | $36.2 | $36.7 | $26.9 | 16.0% | $61.0 | $46.6 | $44.4 | $51.3 | www.macrovision.com |
| 28 | Verint Systems | Monitoring software | $249.8 | $192.7 | $157.8 | 25.8% | $19.1 | $17.9 | $10.1 | 37.5% | N/A | N/A | N/A | N/A | www.verintsystems.com |
| 29 | Diodes | Semiconductors | $214.8 | $185.7 | $136.9 | 25.3% | $33.3 | $25.6 | $10.1 | 81.6% | $61.4 | $54.2 | $50.6 | $48.6 | www.diodes.com |
| 30 | Molecular Devices | Bioanalysis systems | $181.2 | $148.5 | $115.6 | 25.2% | $15.9 | $17.2 | $7.7 | 43.7% | $52.5 | $45.2 | $44.5 | $39.1 | www.moleculardevices.com |
Bill Roberts is an ELECTRONIC BUSINESS contributing writer.
