99% of Last-Mile Delivery Services Will Fail, So Be Careful


When “venture”* capitalists raise larger and larger funds, they write larger and larger checks. At its simplest, “venture” firms have a limited number of partners and cannot possibly deploy large funds with $5M checks. Thus “venture” firms keep the number of deals flat but increase check size… lest they take on more partners.

This has created a swelling of private market valuations in the later stages of “startup”* lifecycles. Larger investor checks = larger valuations = larger desire for growth = subsidized products and services for market share = miserable/unsustainable unit economics. And nothing demonstrates this better than the last-mile, on-demand delivery market.

Few of today’s founders will remember – perhaps by choice – Webvan, a last-mile delivery service from the late 90’s. Similar to today’s last-mile players, Webvan set out to deliver groceries to customers within a 30-minute window. Numbers vary, but Webvan reportedly earned $400,000 in cumulative revenue prior to its 1999 IPO, where it raised $375M.

The rest, you could say, is history. Webvan struggled with user growth, the infrastructure required to successfully deliver groceries – even though it was only in a handful of metro areas, and, above all, the economics to ever get rightside-up.

Investors in today’s last-mile services will argue that today’s market is different. Internet users have increased 10-fold. Commerce is no longer limited to desktop computers but ubiquitous in consumers’ pockets. The cost of running  a startup has dropped precipitously with services like AWS and open development frameworks.

All of these are good points, and without specifics, true. What they neglect to mention is the relative accretive value with on-demand delivery. Most of today’s delivery startups are focused on low value purchases from low margin producers. Grocers average a margin of 2%. Restaurants average a margin of 4%. Thus consumers are conditioned to pay close to true cost for these products.

When a company like Instacart or DoorDash charges me $3 for a $15 purchase, it’s reasonable. I figure how long it would take me to drive to the retailer in traffic, park, go inside, transact, and head back to work. Add up the value of my time (which isn’t much!) and the gas/wear-and-tear on my Subaru (you can use a DOT chart to determine current reimbursement rate) and it’s usually a sound economic decision. But that’s not the true cost to these “startups”. At least in today’s market. Here’s a quick hack to prove it.

Go to TaxiFareFinder to estimate how much it would cost to have a taxi deliver food to you. I took a screenshot of the estimated fare for some pizza to be delivered to a Houston resident 3.5 miles from the pizza shop (note: this is not my address because it’s probably nicer than I can afford).

Now we have to think about margin. The best report I’ve ever seen was a five-year compilation of taxi costs published by the city of San Francisco in 2005. For those taxi operating companies that were profitable, the average margin was 20%; 36% of the operators were unprofitable. If an average transport service charged $15 for delivery, it really wouldn’t be making that much.

So how does a startup like Instacart look at these economics and think, “Awesome”? Using the fare calculator above, if I want my $15 pizza delivered, delivery just doubled the price! Since I’ve been conditioned to pay true cost for these kinds of items, for $30 I’m making an entirely different purchase decision. Just forget about sub-$15 purchases: the notion that I’ll spend $5 on a hamburger and shell out THREE TIMES the meal’s intrinsic value to have it delivered is ludicrous.

One way around this is to focus on high-dollar activities. PayonDelivery focuses on transaction sizes worth more than $100. Their strategy is logical: consumers are not going to pay double what the underlying good is worth to have it delivered. If I’m buying a broken iPhone for $20, I’m not going to pay $40 to have it delivered; for $60 I’m buying a used phone on eBay.

Another solution might be to focus exclusively on dense metro areas where positive economics can be achieved with a smaller number of deliveries. Webvan focused on the San Francisco Bay Area, San Diego, LA, Dallas, and a few other cities… but that did not work out for them.  Further, I’m not sure you can sell investors on a trillion-dollar business admitting you will be limited to dense metro areas with reasonable infrastructure (i.e. > third world) and high average sales prices (i.e. > second world).

“Sure Jordan, anyone can throw stones. Do something useful!”

Fine. Here’s how last-mile makes sense:

  1. Very high supply/demand. When a 200-seat airplane flies from LAX to LAS with one passenger on board, the cost to service that passenger is the entire cost of the trip. When another passenger buys a ticket, the cost to serve each customer slices in half. Assuming passengers are buying tickets at the same price, at some number of passengers the price of a ticket covers that passenger’s cost and earns the airline a profit. To achieve positive economics, the last-mile delivery services will need an on-demand driver to have many packages in his truck at the same time. Whether the food can remain fresh, temperature-appropriate and and be served in a reasonable period of time are certainly debatable.
  2. Massive changes in delivery technology. Moving away from cars and using bike couriers or rickshaw drivers would work… but only in dense metro areas in places like India where you can pay delivery drivers a few cents per day (this is how Zomato scaled their menu-curation business). The elephant in the room is the driverless car, but who knows when this will be a reality for your average American city.

I have a tough time seeing where these last-mile providers get to positive unit economics. 2016 should be the year of transparency, however, as many of the late stage companies in San Francisco run out of funds to fuel their excessive burn; we saw companies like Sidecar go belly up even before 2016. It should get interesting soon enough.

UPDATE: Foodpanda, after raising $300M for food delivery, appears to be shutting down its India operations, despite having 500 employees weeks ago.

*I use quotations around nominal terms which are, in fact, something different. A company earning tens of millions in revenue with 500 employees is not a startup, regardless of EBITDA. An investor who takes little risk likewise does not deserve an association with the word venture.