Startup Forensics: The Fall of Webvan

The Webvan IPO prospectus offered US$375 million gross, then the company closed on 9 July 2001 after further financing failed. Here is why Webvan failed: it built warehouses, stock and delivery capacity for a demand its first market had not yet shown. The figures below show where the model broke.^{[2] [4]}

Our read is that the failure was one of sequencing. Infrastructure and market expansion assumed demand that local operations had not yet established.

The Patient: Webvan Funding and the Promise of Same-Day Groceries

Webvan promised same-day groceries, ordered online and delivered to the home. Its answer was a system of automated warehouses, stock and a delivery fleet, so every order depended on enough volume to pay for it all.

The IPO prospectus offered US$375 million gross. Contemporary reporting documented US$393.4 million of pre-IPO venture investment. Those figures sum to US$768.4 million before offering costs and any additional issuance, which is not a verified lifetime net-funding total.^{[4] [5]}

Benchmark and Sequoia were early venture backers, according to contemporary reporting.^[5]

• The offer: same-day groceries ordered online and delivered to the home.
• The model: automated warehouses, stock and a delivery fleet, which needed enough orders to pay for them.
• Total funding: US$393.4 million pre-IPO and US$375 million gross IPO. Early backers: Benchmark and Sequoia.^{[4] [5]}

The Symptoms: Webvan Warning Signs in the Bay Area Order Gap

The gap showed up in its first market. Reporting put Bay Area break-even demand at about 8,000 orders a day, against roughly 3,000 at the best level the company reached. Those are local, company-reported operating figures, not a global customer acquisition cost.^[1]

The 10-K says the same thing across the network: distribution centers ran at under 40% of designed capacity, against a design of 8,000 orders a day. Idle capacity still carried rent, equipment and staff.^[7]

A warehouse can lower picking cost when it is busy, and an underused one makes each order expensive. Delivery routes depend on density in the same way. Expansion repeats the fixed investment before that relationship is proven.

In April 2001, Webvan reported first-quarter sales of US$77.2 million and a loss of US$86.1 million. The company was conserving capital and integrating HomeGrocer. Revenue was still smaller than the quarterly loss. Those amounts describe one quarter, not the cost of one order.^[6]

• Bay Area break-even demand was about 8,000 orders a day, against roughly 3,000 at the best level reached.^[1]
• April 2001: first-quarter sales of US$77.2 million against a loss of US$86.1 million.^[6]
• The company was conserving capital and integrating HomeGrocer at the time.^[6]

Cause of Death: Why Webvan Failed Before Local Demand Caught Up

The Webvan shutdown came on 9 July 2001 after further financing failed. It planned bankruptcy and laid off about 2,000 workers.^[2]

Our read: the Webvan business model failed on sequencing. Infrastructure and market expansion assumed demand that local operations had not yet established. The collapse in dot-com financing made the stop immediate, but it did not create the demand gap.

Management controlled the commitments and their timing. This evidence does not prove that online grocery delivery itself was impossible. NDTV reported in 2013 that Amazon was building a grocery business from the ashes of Webvan. Later operators could use different infrastructure and market-density choices.^[3]

The Toxicology Screen: The Numbers Behind Webvan's Collapse

The US$375 million was the prospectus's initial gross IPO size. Underwriting and later option exercises affect what the company actually received. The US$1 billion warehouse-construction commitment should not be confused with capital raised, because a planned expenditure is not money in hand.^[4]

The annual report gives the cash picture. Webvan lost US$453.3 million in 2000 on net sales of US$178.5 million, and its accumulated deficit reached US$612.7 million. Net cash used in operating activities was US$263.1 million for the year. That is an annual figure, so the Webvan burn rate per month stays unknown. Cash and equivalents were US$40.3 million at 31 December 2000, and the auditor raised going-concern doubt. No runway figure was published.^[7]

Pricing is only partly visible. Delivery was free above a minimum order size, which Webvan raised to cut costs, but the amount is not stated. Average order was US$104 in 2000. Acquisition cost and churn were not published.^[7]

• Customer acquisition cost (CAC): Not publicly disclosed. Order-level contribution is also not disclosed.
• User churn: Not publicly disclosed.
• Pricing model: Free delivery above a minimum order size, raised to cut costs; amount not stated. Average order was US$104 in 2000.^[7]
• Monthly cash burn: Not published monthly. Net cash used in operating activities was US$263.1 million for 2000, an annual figure.^[7]
• Runway at collapse: Not publicly disclosed. Cash and equivalents were US$40.3 million at 31 December 2000, with going-concern doubt.^[7]
• Total funding: US$768.4 million, pre-IPO plus IPO gross, documented. Not a verified net total.^{[4] [5]}
• Reported loss: Net loss of US$453.3 million for 2000; accumulated deficit US$612.7 million. First quarter 2001 loss was US$86.1 million.^{[6] [7]}
• Revenue: Net sales were US$178.5 million in 2000, up from US$13.3 million in 1999.^[7]

Open full-size image

Exhibit A. Figures, periods and uncertainty follow the cited records.

The Post-Mortem: Lessons for Founders

Prove one delivery zone before a second lease. Require repeat purchases and contribution after picking and transport before you sign a second warehouse lease. Report break-even orders and actual orders on the same local dashboard. Slower expansion gives rivals time, but that is cheaper than multiplying a demand gap.

Keep reversible spend separate from hard-to-exit commitments. Divide the cash plan into reversible operating spend and commitments that are hard to exit. Test six months of slower orders with no new financing, then approve construction in phases. Let customer convenience earn each expansion through paid usage, not press attention.

Further reading

• Webvan IPO prospectus, 5 November 1999; indexed filing excerpt
• SFGate, “Webvan goes under,” 9 July 2001
• SFGate: Webvan’s efforts to keep on truckin’, 27 April 2001

More Startup Forensics teardowns: Why Theranos failed · Why Quibi failed · Why Jawbone failed

Want a teardown like this every week? Subscribe to the Startup Forensics newsletter, or leave a comment telling us which startup we should investigate next.