Clarify:

Since Uber has different products, can we just assume Uber Rides? Sure

Also do we want to consider all uber drivers or XL, luxury segment separately? Include all of them.

Ok let me start.

The US population is 300M

There are 50 states therefore avg is 60M.

As far as I know NY has 20M population. Let me take 1/2 of the NY for a med sized city.

Population of a med size city: 10M.

If each household has avg 3 members: around 4M households.

Assuming each house has 1 vehicle: 4M vehicles in a mid size city

What could be these vehicles?

Very old cars

Cars with 4 doors

Truck

Van

Mini bus etc

To be an Uber driver, you need to show the proof that your car has 4 doors and min 3 passengers can use it. In addition very old cars cannot pass the assessment.

If this city is a mid west city, I assume that ppl are using the trucks and big vans more than the metropols. Therefore, I assume that the  trucks, big cars etc  vs normal car would be 1:4 therefore

We can assume that there would be around 3M available cars for Uber.

I assume there can be 1 driver from 1 household therefore 3M potential drivers.

I need to eliminate if ppl has some disabilities or hearing, sight loss etc and also not wiling to work at Uber, or their work does not let this to do, we need to reduce this from the potential. I will cut around 20% from the volume.

3M*0.8 = 240,000

The competition in the city between Uber and Lyft : 65% vs 35%

160,000 potential uber drivers

Ppl use Uber more in the city centers and they do not prefer it at suburbs. Since any mid-west city’s city center population could be around 1%. I will take the 1% of potential uber drivers in the city.

Around 160K*1%= 1600 potential drivers for Uber in a mid west city with 10 M population

Assumptions:

• A medium-sized US city has a population of 250k citizens.
• One person per ride (median).

   

• The demand and supply balanced the market, so Uber drivers are fully occupied.
• In a typical US city, Uber will own up to 50% of the market share.
• I’m going to make my calculations in the following way:
  • The calculation will be made for a rush hour; if Uber could handle it, it could handle the rest of the day as well. (bottom-up approach).
  • Calculate how many people will use the service in rush hour.
  • How much time it takes to serve a person.
  • How many peoples a driver can serve.
• The final result will be = 50% X [# of people to serve during the rush hour] / [How many people a car can serve during an hour]

[# of people to serve during the rush hour] = Let’s say the rush hour will be between 16-17 when two significant events occur:

1. School College University finishes
2. A workday finishes (there is a % of workplaces who are also starting the 2nd watch)

How many students will use a ride-sharing service?

• Assumption: Most students study close to their homes (by foot, bike cycle, scooter,…), use school buses, have vehicles, drive with friends…
• Therefore, I assume that the audience is about 10% of 15-25 population.
• Assumption: US life span is 80 years old with uniform distribution, -> 1/8 for every decade -> ~30K at the age of 15-25
• 10% -> 3K students that might use the ride-sharing services.

Workers

• Age 20-60 -> 4 decades -> 125K (half of the population)
• I assume that most of them own a vehicle, work close to home, use transportation, or other alternatives.
• I assume that only 5% of them consider a ride-sharing service -> ~6K
• Total people to serve ~9k during rush hours

How much time does it take to serve a person

• In a typical city, I assume that it takes:
  • 10min to get anywhere (during rush hours, since we have optimization algorithms)
  • 2min to pick up and pick down the rider
  • 1min to breath focus get the next order
  • In total 13min
  • So during an hour, a driver can serve about 5 people.

To serve 9,000 people during rush hours, we need 9,000/5 -> 1,800 drivers

Final calculation result, # of Uber car drivers= 1,800 X 50% -> 900 drivers

Sanity check,

Households in the city 250,000/3 -> 83,000

It means that 2 percent of the households have a ride-sharing driver? I think that it makes sense.

Please give me a moment to think through some of the factors involved and then come up with a formula.

• How many people are taking trips
• How many people are in each car.
• • Take into account pool vs uber x
• Optimal average wait time
• • How long a trip is.
  • How long it takes driver to get to next trip

Okay, so from here, I want to set up a formula, and then I’ll deep dive into each of these to determine the values.

Number of drivers = (1) number of people that want to take a trip in a certain time period) / (2) people per car.  (3) Then adjusting for some wait time factor.

So, let’s get started.

(1) Number of people that want to take a trip during some time period. 

So we said we’re in a medium sized city. So let’s say that’s 1M people.  Now we need to figure out how often people are taking trips.

So first question is how many people in a city use Uber. A few factors to consider:

• Children are generally not taking Uber, and elderly people are generally not taking it either. We also know that Uber has competition in most medium-sized American cities (i.e from Lyft).
• We also know that in most medium-sized cities that people have cars and that’s their main type of transportation.
• So the main use case for Uber is more if they don’t have a car and/or they don’t want to use their car (e.g. they’re going out for the night and they’re drinking).

Given all of that,  20% of the population using Uber seems reasonable.

So that gives us 200,000 people using Uber regularly in a moderately sized city.

Next we have to figure out how many trips people are taking during some appropriate time period.  I think people are going to be using Uber more on the weekend when they’re going out, particularly Friday and Saturday nights. I know that Taxis and Ubers are most busy then. So, I want to start with the required Ubers during that peak period.

So let’s say that on a given Friday and Saturday night 5% of Uber users will be using Uber (1 in 20 Uber customers are using it. That seems reasonable, maybe even a bit high).  So that gives us 5% * 200,000 = 10,000 people using Uber on a given week-end night (Friday/Saturday night). But each of those people need to go there and back. So let’s say that’s 20,000 trips on a Friday/Sat night. This seems like a lot. So may need to go back and adjust this assumption. 

And let’s say Friday/Saturday night lasts for roughly 6 hour period. So that’s roughly 20,000 / 6 = approx 3,000 people per hour on a Friday/Saturday evening

(2) Now, how many people per car?

• Pool (3 people per ride): 50% of rides
• UberX (Let’s say 1.5 people per ride): 50 of the time
• Gives us weighted average of 1.5 + 0.75 = 2.25 which approx 2 people per ride.

So that gives us 1,500 trips per hour on a Friday/Saturday evening.

(3) Okay, so now we get to wait time. 

How long is the average drive.  In a moderately sized city, let’s say that 20 minutes.

We also need to take into account that drivers need to get to their next destination, so let’s add another 10 minutes onto that to get to the next trip. Sometimes, if they’re driving to a busy commercial area, the next trip will be very close, but other times it’ll be far. But let’s say that 1/3 of drivers’ time is spent getting to a trip, and 2/3 is spent driving, so that would be 20 minute trips plus 10 minutes to get there. That seems reasonable.

So if we wanted a 30 minute wait time, we’d need 1,500 (trips per hour) x 3 = 4,500 cars (drivers) available every 30 minutes.  However, that’s too long. Let’s say Uber wants to ensure an average wait time of 10 minutes during that peak period.  So then we need triple that amount of drivers. Which is approx 15,000 drivers in order to ensure a 10 minute wait time during the peak period of Friday/Sat evening.

Refections / self-feedback

This seems too high because that’s 1 driver for every 66 people in the city.

Some of the major assumptions were

• 1M residents in a moderately sized city
• 20% of people that use Uber
• 5% using it during peak period
• 10 minute acceptable wait time

I was able to find out that Lyft (which is similar enough to Uber) has 1.4 million drivers in 300 US cities, so that would be ~4,500. However, that’s for the average city, so I think 1M is bigger than the average city. After a Google search, I realized that average city is probably more like 300,000 people. E.g. Pittsburgh is considered a midsized city and its population is 300,000 people. So my incorrect assumption was that there was 1M rather than 300,000 people in mid sized city. If I divide my starting population by 3, then that will give me 5,000 drivers, which is very close to the Lyft number of 4,500 per city.

I also should have asked at the beginning whether we’re calculating for a mid-sized metro area or just for the city. E.g. a mid-sized metro area is probably more like 1-2 million. But a mid-sized city, is more like 300,000 (e.g. Pittsburgh’s city population is 300k but its metro area is 2M).