Why Dot-Com Disaster Kozmo Never Became Instacart

An investigation two decades later

Where Are They Now is a column that revisits once-popular companies and brands that have seemingly disappeared.

The shorthand for the bubbliest startup of the dot-com bubble has long been Kozmo, the turn-of-the-century startup that intended to solve the ultimate logistical problem: What if someone would just bring me that thing I want, now? Imagine Amazon Prime at the speed of pizza delivery but for free. At the time, many people asked “How is that a feasible business model?” — including the founders of Kozmo, though not soon enough to save their IPO in 2000.

And yet today, the world looks a lot like Kozmo envisioned it, especially during the pandemic. A network of companies — some old, some startups — have collectively pulled off something close to Kozmo’s original vision. And this time, the services seem to be more than sticking. DoorDash’s stock reached more than $189 a share on its first day of trading, a $72 billion valuation, and it’s now at over $200 a share some two months later. Instacart is bulking up its corporate team as the market waits for an IPO. Pandemic conditions caused Amazon to hire over 400,000 new employees in the first 10 months of 2020. Sales at Target’s in-house delivery service, Shipt, grew 300% over the holidays while Uber recently acquired liquor delivery service Drizly for $1.1 billion. So what stopped Kozmo from becoming Instacart?

Kozmo’s founder, a 25-year-old Goldman Sachs investment banker named Joseph Park, started the company in 1997 after realizing he would have to wait three to five days to read a John Grisham novel if he purchased it on Amazon. He named the company after his favorite drink and favorite Seinfeld character (cosmopolitan, Cosmo Kramer) and plowed a fifth of his $100,000-a-year salary into it. His freshman college roommate, Yong Kang, signed up to be the company’s co-founder, and the pair raised $232 million by the time Kozmo filed for its ill-fated IPO almost three years later.

The challenge for Kozmo, which operated in 11 cities at its peak, was balancing its cartoonish elevator pitch (“Order anything from our website, and we’ll deliver it to you for free within an hour”) with the logistical precision to sustain it. While the narrative was instant indulgence, the reality was using delivery to replace a big-box store. Or, well, a small-box store, something smaller than a Target but bigger than a 7-Eleven. Instead of delivering stuff to a warehouse to be delivered to a retail location to be sold by staff, orders were delivered by bike from leased warehouse space. Some of the items Kozmo bought, trying to predict what its customers wanted; some it took on consignment, with the financial risk borne by the manufacturer, a split that looked a bit like Amazon’s current warehouse-and-third-party model; and some were in-between promotions, splitting the risk.

Putting items together into orders and getting those orders to customers in the era of candy-bar phones required building bespoke software and hardware. “Today, you can start up a website on Shopify, and it has an inventory management system built in,” says John C. Wu, who was Kozmo’s director of logistics operations and is now president of underwear company Tommy John. “There wasn’t an e-commerce platform for us to use; we had to build that too. We had to build software to maximize the amount of productivity that we’d get out of the couriers. A lot of the money we raised was used to build systems.”

Kozmo wasn’t an everything store like Amazon. It didn’t carry older, niche books, but it did have, say, the latest John Grisham novel, one of 300-some titles on offer. It had 14,000 movies, determined in part by deals it struck with studios. It had an inventory of about 2,000 CDs, 200 magazines, medicine cabinet basics, smokes, snacks, ice cream. Before streaming and e-books, if you wanted the most popular thing, you could go to your local big-box, and they’d probably have it, and if that wasn’t what you wanted, they’d probably have something good enough — plus snacks and some basic household supplies you could pick up while you were there.

That was Kozmo’s sweet spot: a department store run. Unfortunately, it also wanted to serve the person who pops in for a candy bar. And if you tell the internet, even the fledgling ’90s internet, that you will deliver them a candy bar at retail cost for no charge, the internet will take you up on that. The average order size in January 2000 was a mere $5, more like a robust convenience store stop than a Target run.

“Our core customers were single people between 20 and 34 with low disposable income,” Wu wrote in a thorough 2004 postmortem in Supply Chain Management. “They were using Kozmo to order a video and a snack, to get a meal delivered, to buy low-cost blank CDs on which to burn their Napster downloads.” The cost to Kozmo for each order was $7.50.

Labor costs were also higher for Kozmo than for current iterations of on-demand delivery companies. Prior to the gig economy model — which took years of hardware and software evolution — Kozmo had to compete with courier companies for delivery staff. So instead of pinging gig employees when a delivery was ready, it trained its couriers in warehouse restocking and order picking and packing. That way the company had them around when it needed them for delivery surges, and they still had things to do in the meantime.

Kozmo scrambled to increase its customers’ order size. It killed its dream of universal free delivery, instituting a $2.50 fee for orders under $30. It tried to target the big-boxes’ best customers, dual-income families and parents, with high-margin “packages” that encouraged shoppers to buy a bulk order of baby stuff in one go. And it worked; the average order size increased throughout the year 2000, reaching $40 in January 2001. In February 2000, Kozmo partnered with Starbucks. Amazon, the company Park wanted to catch and surpass, invested $60 million in March 2000 and announced a three-year partnership in which Kozmo would sell its frenemy’s products. In July, which would prove to be its peak, Kozmo employed 2,665 people.

In December 2000, Kozmo turned a profit in New York, San Francisco, and Boston. (The one city Wu describes as a “lost cause” was Los Angeles because its sprawl made orders harder to cluster and because its car dependence forced Kozmo to buy vans.) Kozmo’s average order size in January 2001, when everything started falling apart, was $40. Despite these signs of a sustainable business, Kozmo had expanded too quickly and spent too much to survive the popping of the bubble that it, in part, defined. Having burned through more than $200 million, their planned March 2000 IPO was delayed until August 2000, then pulled altogether. In April 2001, almost exactly four years after it was founded, Kozmo came to an end.

But on the floor, Kozmo’s young team carved out an efficient logistical operation with tech tools and a labor model largely built from the ground up. And for a brief time, the vanguard of web 1.0 delivered a glimpse 20 years in the future.

Freelance writer/editor in Chicago. Words in Marker, The Atlantic, COVID Tracking Project, elsewhere. Author of ‘Chicago: From Vision to Metropolis.’

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store