Why did Webvan fail so spectacularly?
Webvan was established in 1997 to allow consumers to order groceries online and have them delivered to their home within a 30 minute window. It attracted hundreds of millions of dollars in venture capital and hundreds of millions more in a 1999 IPO because investors believed that the company could use online grocery sales as a spring board into other business lines. The high priced, well-decorated management team at Webvan wooed investors with promise of solving the “last mile” problem, how to bring goods directly into customers’ homes profitably. Why this group of top-talent executives believed they could not only solve the near mythical problem that has long plagued retailers but also solve it on a scale large enough to service markets all over the county in two years is beyond me. Webvan might have been successful if it had attempted to service one market efficiently and then expand into a new market using the knowledge gained by efficiently serving the first. Instead, Webvan built a complex distribution center that management believed was capable of being replicated quickly in many new markets. Besides investing hundreds of millions of dollars to support a market that did not yet exist, three fatal errors doomed the company’s model.
Webvan created its distribution center around a maze of conveyor belts, scanners, and picking stations. Each new distribution center costs $35 million just to build. Stocking the center and marketing the company’s service in a new market, and staffing added to that total. A significant part of that cost was tied to the elaborate carrousels that Webvan kept its products in. The carrousels rotated products around the employees so that they could remain stationary and fill orders faster. However, this technology only helped fill about 35% of Webvan’s orders because it could not store produce of frozen goods, the remaining 65% of ordered still required employees to manually select products. The investment in the carrousels and the technology to make them work failed to create as much of a cost savings as management had envisioned.
Another obstacle was delivery density. Webvan needed 10-12% of total households to be repeat customers for its model to be profitable. In 1999, that might have been a half to two-thirds of the tech-savvy households in the region. If all of those households were clustered together in one or two neighborhoods, Webvan could have serviced them with a significantly less capital intensive model. Perhaps a smaller, more centrally located (with respect to customers) distribution center with only a small fleet of vans. Unfortunately for Webvan, in 1999 the tech-savvy households that they needed to count as repeat customers were scattered all over major urban/ suburban areas and required a massive operation to provide the level of service that Webvan claimed to be provide.
Too Smart for Their Own Good:
Webvan assembled a who’s who list of top talent executives. They created data and models that supported their own beliefs and when analysts and investors voiced concerns about Webvan’s assumptions, management just shrugged it off. The Founder, Louis Borders proclaimed “Its $10 billion or zero” and CEO George Shaheen believed that Webvan would be more than groceries and that the market was $1.5 trillion (total projected online purchases for 2003 by IDC) not $6.5 billion projected for online grocery sales by Jupiter Communications for the same year. What is truly amazing is not so much the projectitis that this group contracted but that so many analysts and investors were sipping the Kool-Aid too. It’s not uncommon for managers to get so far caught up in their own ideas that they fail recognize what’s actually occurring in the market but this was star-studded management team who had been there, done that, and knew to question every assumption relentlessly to be sure that it was reasonable. Unfortunately, this management team believed that they were different. They had the best minds and the best technology available. They believed their solution was better than everyone else’s who had tried and failed to solve the “last mile” problem before them. They casually shrugged off conflicting data and ignored analysts’ concerns to their own detriment. In the end, I suppose that if you are going to fail, you might as well be the best at that too.