28 September, 2015

Uber (Part 5) - Unit economics of food delivery

Relative to Uber's other challenges, food delivery seems like an expensive distraction unless their experiments in food delivery convince them to buy/partner with GrubHub.

A Sidecar blog post recently quoted Fred Smith, the CEO of FedEx:
“I think there’s just an urban mythology out there that an [on-demand delivery app] somehow changes the basic cost input of the logistics business or changes the patterns or the underlying business situation and that’s just not–that’s just incorrect. So great company, great concept [in reference to Uber], but I don’t think it’s…likely to be a major player in the logistics business”

The informative post then detailed the numbers behind Sidecar's multi-modal delivery system and posited that their system, when fully loaded, could deliver a package anywhere in the city for about $6.90 using San Francisco labor economics, less than half the cost of a FedEx $16 2-hour delivery:

  • Cars can make 6 dropoffs/hour if they don't have to park and get out; drivers need to make about $22/hr 
  • Bikers can make 4 dropoffs/hour and riders need to make about $19/hr
  • Walkers need to make about $16/hr

UberEats appears to be pricing food delivery well below run-rate cost

Taking Sidecar's numbers at face value, it's an interesting localized disruption of bike messengers and FedEx.  But the $7/package is what's most interesting to me - compare it to $8 for a one-hour delivery in Amazon's PrimeNow, a service that's doing its pickups at a central warehouse.  Note that Amazon also has underlying retail margins which help "pay" for delivery (a central recognition of Prime).

One of Uber's ideas for how its platform of on-demand cars could be extended is food delivery.  In August, it launched UberEats in SanFrancisco, offering lunch in a limited area for a $3 delivery fee.

While Uber is likely somewhat subsidizing food delivery in order to learn about the business, the difference between $3/order and $7-8/order is the difference between what makes or breaks aggregated food delivery from local restaurants.

It's possible that Uber believes it can utilize its cars alongside food delivery, like an UberPool except riders are sharing the vehicle with food.  It would be interesting to imagine what combination of dynamic hand-offs and passenger delays (both food and people) would make this successful.

Two problems with food delivery 1.0 as a business: Multiple pickup locations and managing vendor relationships increase cost of intermediation

Local restaurant food delivery, where available, works by paying very low wages to people with "beater" cars to make delivery loops and earn (limited) money through tips.  If a driver can drop off three orders in 30 minutes, there's a reasonable chance that customers can get mostly-fresh food within about an hour of ordering.

Restaurants run on very low food margins; there's not much room to "pay" a third party for order aggregation, and locals tend to know who they like to eat from.  Often, delivery drivers will perform chores at the restaurant during slow hours.  Note that (in PA at least), higher margins from recently allowed alcohol delivery may help somewhat.

There are two issues with traditional food delivery rollups that make the unit economics challenging.

The first issue is multiple pickup locations outweigh any benefits from "network effects" of delivery.  The second issue is that managing relationships (menu, QA, order coordination, etc) with multiple local restaurants is a costly and hyper-local thing.  This second reason is why Groupon's cost structure never achieved scale (separate from limits to Groupon's value proposition to local businesses).

The success of GrubHub (which merged with Seamless in May 2013) as a delivery rollup has encouraged others to attempt to compete despite those two issues.  Here's a review of seven of them in Seattle, last year.  It noted some of the challenges of intermediating customer service.

GrubHub appears profitable - about $16m profit off $88m in sales in the quarter ending June 30th - and growing (50% revenue growth from same quarter in 2014).  Note that GrubHub is currently trading around $24/share, down from $47 in April 2015.  GrubHub's market cap of $2b today is comparable to Groupon's of $2.4b, offhand I don't know how comparable their footprints are.

A quick internet search suggests that until recently GrubHub didn't actually deliver - it was a front-end ordering system only, and left restaurants to fulfill their own orders.  Presumably restaurants believe there is upside to paying GrubHub to bring in orders it would otherwise miss out on; also presumably restaurants will benefit from competing ordering platforms lowering the fee paid to GrubHub.

It's beyond the scope of this post to examine how GrubHub intends to solve the delivery issue, but its 50% share price drop over the last few months likely reflects investor perception on the impact of increased competition and added delivery costs on GrubHub's growth trajectory and margin retention.

Food delivery 2.0 exists, but still seems hyper-local with few economies of scale or network effects

A second type of online food delivery has emerged, one that delivers its own food.  Muchery, Lisch, and Spoonrocket deliver food made by their own in-house chefs.  This essentially solves the two issues identified above that plague traditional food delivery startups.

At first glance, these own-cooked delivery services don't seem significantly different from a traditional restaurant, except perhaps rent savings from non-prime real estate and having to pay for table space.  Rent is about 5-8% of a restaurant's cost structure, and total occupancy costs no more than 10%.

At second glance, these own-cooked delivery services are executing the food-truck model of restaurant cost arbitrage, except adding a costly layer of delivery in exchange for extending their reach.  I could also imagine a hybrid of these two, with a streetside table setup below a cheap upper-floor kitchen, and runners taking orders from the kitchen and "delivering" them to the street as well as handing them off to delivery drivers.

There are many reasons why restaurants, as a category, fail to return their cost of capital.  One is that they are passion investments.  The two models which seem successful are franchises, where the risks are born by the local investor and the upside collected by the central franchise; and alcohol sales, which have much better margins than food.


The run-rate economic cost of micro-delivery appears to be higher than what can be supported by food delivery, even when mixed with other types of items.  The $7/order cost is about twice what Uber is currently charging even before considering customer support costs for both users and restaurants.

While cars could play a role in the broader local delivery market, it's more likely that they would be as an "on-demand" addition to a network such as Sidecar's.  Which is, itself, very labor intensive.

There doesn't appear to be much value from an Uber partnership with Sidecar; the value would accrue to Sidecar and anyway there is likely much overlap in the peak demand usage patterns.  An "UberPool" with food model seems better, but the costs of intermediating restaurant orders seem high relative to the margins involved.

Questionable run-rate economics of food delivery could be improved by an increased prevalence of UberPool, particularly if the platform orchestrated vehicle transfers by its passengers.  Relative to Uber's other challenges, food delivery seems like an expensive distraction, unless their experiments in food delivery convince them to buy/partner with GrubHub.  Because Uber's stock appears overvalued, this would likely need to be a cash offer - at $2b, a GrubHub acquisition would be within Uber's reach but maybe not the best use of the money.

The end game in hyper-local delivery seems like a combination of right-sized autonomous vehicles with enough baseload demand capacity from their own sales.  Like a fleet of autonomous Amazon Prime drones.

Thanks for reading,

1 comment:

Focused, topical critiques are welcome though my responses may be limited. Off-topic comments, trolling, spam, or anything not pulling its weight will be deleted.