The Economics of Same-Day Delivery

I recently needed a duffel bag on short notice before a trip. Usually, I would make a trip to REI or Sports Basement, but instead, I decided to try the service UniHop, which showed up when I googled "REI Same-Day Delivery." For $15, I can hire a driver to pick up my order (placed on rei.com) and drive to Stanford. This is an interesting price—often we cringe at high delivery fees for DoorDash orders ("how is my burrito $22?!") but actually, $15 for this trip seems more than reasonable. I wanted to break down the opportunity cost to see exactly how much convenience actually costs.

First, let's think about the actual drive. At around 5 PM (when I have the trip scheduled), Google says it will take between 22-40 minutes to drive to REI. Anecdotally, this trip usually takes about 30 minutes at rush hour, so let's use that as an assumption. The trip is 13.9 miles of mostly highway driving. Assuming we wait about 10 minutes for the order to be brought out of the store, this trip would take around 1 hour and 10 minutes, with a total driving distance of 27.8 miles. The driving time would be around 1 hour, meaning an average speed of 28 miles per hour. This is slow enough that I can't assume very good gas mileage. My Subaru Forester with a THULE box on top will probably get about 20 mpg, meaning the trip would use 1.4 gallons of gas. In Palo Alto, where gas is around $4.20 a gallon, the trip would cost $5.88 in gas alone. We also have to take into account the wear on my car. Most calculators estimate the cost of wear at around $0.19 per mile for my car. This means that a 28-mile trip would cost an additional $5.32. Adding this up, we see a total trip "cost" of $11.20.

So for a $14.99 trip, the comparative "cost" is around $3.79. In my opinion, my time is definitely worth more than $3.45 an hour, so this makes the service an easy sell. In fact, it's such a good deal that I would consider doing most of my shopping this way! This calculation also fails to consider other externalities that come from making the trip to REI: driving is dangerous, and there's a small chance I could get into an accident or otherwise be injured in some way. In the US, there are around 80 injuries per 100 million miles traveled, leading to a 0.00224% chance of injury on this trip. If the weighted average cost for possible injury is around $110,000 (according to 2023 NSC data) and we account for the fatality rate (1.26 fatalities per 100 million miles) and cost per fatality (about $13 million), then the possibility of injury costs us about $0.17 per mile. Since highway driving has a lower crash rate, we can probably lower this to somewhere between $0.10 per mile and $0.15 per mile. Taking the conservative estimate of $0.10 per mile, the possibility of injury adds around $2.76 to this trip, making it almost break even. Note that this is back-of-a-napkin math, but most sources estimate the possibility of injury/accident at over $0.10 per mile.

These calculations, of course, shine a light on the cost of gig work and whether it really is beneficial for the economy. There is a long-running debate over whether Uber/DoorDash is actually profitable for the drivers, which I am not qualified to weigh in on. But actually doing these calculations makes it hard for me to work out how these drivers could ever break even. Maybe my burrito should be much more than $22...