Here’s an example of how you might solve a typical tech case interview: How much will a self-driving Uber ride cost?
So Uber has launched self-driving service in Pittsburgh. And while the sci-fi fan in me sees this as one more step towards the Robopocalypse (written by a Carnegie Mellon robotics PhD, naturally!), the Excel nerd in me sees this this as an awesome opportunity to bust out a spreadsheet.
Namely, could we do a little back-of-the-envelope math and predict how much these rides will actually cost at scale?
So let's start with this basic formula:
Price = Cost/Mile x Miles/Trip x (1 + Margin)
Note that this strips away the ambiguities of Uber's existing consumer-facing pricing model (Base Price + Time + Distance + Booking Fee). Because we don't know how they arrived at those calculations, it makes sense to focus on Uber as a regular business, plain and simple. In other words:
They need to cover their costs
They need to make a profit
Of course, this assumption also removes the venture capital subsidies that have allowed them to operate at a loss for all these years (and, in fact, Uber plans to give away the self-driving rides for free at first). But at some point even VCs get impatient, so let's focus on a point in time where Uber has to stand on its own as a public company.
OK, all that said, let's plug in some current assumptions for each part of our formula:
Cost/Mile = $0.58
This comes straight out of the Department of Transportation's latest average cost per mile for vehicle ownership. And it includes both variable costs (fuel, maintenance, and tires) as well as fixed (insurance, registration, taxes, depreciation, and finance charges). Note: I'm assuming self-driving technology gets within the range of the average vehicle cost - which isn't unreasonable given that even Google's fancy set-up only adds an extra $10K today.
Now, of course, the big thing it doesn't include is a labor cost. So already the cost/mile is nearly halved from Uber's current per-mile charge in Pittsburgh ($1.05 for UberX).
As for how many miles your trip will cover, let's use the most recent national average from SherpaShare (an organization that helps Uber drivers track their mileage):
Miles/Trip = 6.4
Or, to put it in a Pittsburgh perspective, that's a little further than going from Carnegie Mellon (where Uber poached its top robotics researchers from) to the site of the bloody Homestead strike (where Uber drivers might very well stage the next big labor protest!).
When it comes time to think about margin, let's stick with Uber's take on existing fares:
Margin = 20%
Of course, this isn't pure profit as it has to cover all the infrastructure beyond the car itself (customer acquisition, technology development, etc.). But with Uber already a household name and with a solid platform already established, these costs may diminish in the years to come.
So where does that leave us in terms of total price to you, the rider?
$0.58/Mile x 6.4 Miles/Trip x (1 + 20%) = $4.45
Compare that to a current trip of that length in Pittsburgh and you'll see that even the cheapest fare with no surge pricing is nearly 3x as expensive!
OK, so that's pretty impressive. But could we do even better? Let's throw some additional future assumptions into the mix:
Uber's fleet goes all-electric.
Insurance premiums are halved as self-driving cars produce fewer accidents.
Surge pricing is eliminated as Uber no longer needs to incentivize humans to drive at undesirable times.
11.2 cents/mile (or nearly 20%) of the government's forecasted operating costs comes from gasoline. Meanwhile, the average cost of electricity in the US is only 10 cents/kWh.
So assuming Uber's vehicles could achieve 3 miles/kWh, that slashes energy costs to just 3.3 cents/mile. And that takes our average trip down to:
Not only is fuel costly, but so is insurance. In fact, it currently accounts for about 20% of the total cost of ownership for a typical electric vehicle.
So if we could cut those premiums in half based on fewer accidents, we could shave another 10% off our cost structure. Which would drop our average trip down to:
Death to the Surge
Let's go back to that base scenario of an $11 Uber ride across 6.4 miles of Pittsburgh. And now let's imagine that you're taking it on New Year's Eve. In that case, what started as a moderately priced affair might easily become a $30+ investment. After all, how many drivers want to be on the road after midnight on the drunkest night of the year with passengers vomiting all over their ride? Hence, surge pricing.
Whereas, no matter the holiday, the weather, or the level of passenger inebriation, Uber can match its fleet of robo-chauffeurs to the level of demand instantaneously. So, ultimately, we may be talking about a price reduction of not just 60% (as in the standard model) but actually 90%!
At least until all those automated chauffeurs start to lead the Robopocalypse, of course... :)