I am assuming Uber marketplace is where drivers are matched to riders. I would ask a clarifying question on what are the main metrics Uber is optimizing for.

There are following metrics to optimize for: satisfaction, growth, revenue and efficiency

Given Uber’s current situation I would focus on revenue and growth

Listing out all possible metrics:

• Match time – time it takes to match a rider to a driver tradeoff: measures how fast the platform is
• Match accept rate – percentage of first suggestions being accepted ( measures efficiency)
• Match rejection rate from driver (metric will help to make the platform more efficient)
• Match rejection rate from the rider (metric will help to make the platform more efficient)
• Revenue per ride ( how much money Uber is making per ride)
• Rider attrition rate (how many UberX are leaving Uber)
• Number of successful matches per day ( measures growth but might a too vague of a metric, better metric is new uberX users per day)
• Number of new UberX matches per day (measures growth)
• Rider retention rate (how many use UberX again)
• UberX customer lifetime value ( how much money can you make from one user)

Question is to assess how Uber X is doing overall. Can we assume for a specific market, but I am guessing it doesnt matter a whole lot, as the model is more or less similar across markets. Do we need to consider in a non covid world, with no specific mentions of covid related complexities? Also clarifying Uber X is essentially where drivers have private rides versus uber share which are shared rides.

Let us start off with understanding Users and Stakeholders for Uber X and what they are looking for:

Then let us look at goal for Uber X within Uber and how it helps Uber

Finally we can look at the user journey and define some metrics that we will measure, and perhaps pick a NS metric for uber x performance.

Users: Uber X drivers : Drivers own the car, and drive either only Uber and or Lyft. Uber driver can as I understand choose both shared rides or Uber X based on the ride request

Passengers: Are passengers who basically need a private ride. Mostly as they are travelling with upto 4 other people , so friends or coworkers or family, or else alone. so 3 scenarios

1. With other people you already know
2. Alone as want to save time
3. Alone as prefer the privacy
   1. Self paid
   2. Company paid

Uber: Makes a txn % on each ride, and pays the riders for each ride based on certain formula fixed pay + ride commissions.

Goal for Uber X: Engagement and Retention , Reviews, getting existing users to user Uber. Uber is easier , more immediately available versus self driving, public transport options, getting another taxi etc. and that means large # of drivers etc.

User Journey for Uber: Driver signed up with a car -> Passenger wants to travel and searches a ride -> Passenger initiates a request for a ride -> Driver accepts the ride -> Ride completed -> Review for the driver

if a ride is available soon, and the price is right for the ride. So its important to have more drivers in routes so that peak pricing is not impacted. Important to incentivise users to take the UberX versus even share or others.

Supply-demand equation, plus distance.

So higher supply is good, but too high a supply would lead to drivers losing interest as well. So optimal supply demand match.

Indidcated by a metric that makes user finally initiate a ride.

North Star Metric:

# of uber x rides completed (daily, weekly, monthly)

Geographic distribution

Could lead to oversupply and instead of share rides, people are just taking uber x rides. So drivers make less profits price is driven down.

# of uber share rides completed

Ensure it is not cannibalisation

Now if better features exist: maybe cannibalisation isn’t bad.

So need to look at revenue and profitability for Uber

Hence also look at for Uber drivers, average income per ride

# of active uber drivers (daily, weekly, monthly)

Driver Churn

So in conclusion:

We know that UberX is doing well if we get the below NS metric increasing

# of uber share rides completed

And do not see any decrease in the below average income per ride

# of active uber drivers (daily, weekly, monthly)

Driver Churn # of uber share rides completed

Clarifying Questions:

• What is UberX?
  • My assumed answer: getting own private driver door-to-door; the ride is not shared with any strangers (i.e. Uber Pool)

Given the prompt, I would break my answer into two pieces to address to the two parts of the prompt: “describe” and “assess”

Description of Uber Marketplace

• My understanding of the Uber Marketplace, given the context of the question, is that it is essentially a supply/demand algorithm that dynamically matches and prices rides to optimize the wait time for riders given a supply level of drivers with the demand level of riders
  • For example, if there is a low supply of drivers and a high demand from riders, the Uber Marketplace will increase the price for riders to lower rider demand to the point that the wait time for rides will be reasonable. Otherwise, if the Uber Marketplace did not dynamically increase the price, wait times could be an unreasonable time (e.g. like an hour) until a rider gets matched with a driver
  • In addition, the increased price will incentivize drivers to join the network, thus increasing supply and lowering price as well.
• Ultimately, the Uber Marketplace algo is the backbone of the company and the primary value-prop that other rideshare companies compete on
  • Having an efficient/effective pricing model will drive a meaningful and value-add experience for both the supply side and demand side

Assessment of Uber Marketplace

• I would have the following success metric buckets to assess how successful Uber Marketplace is for UberX
  • Cancellations
    • Supply side
      • If cancellations from drivers are low, they feel like they are getting paid a fare/fair (pun intended) amount for each ride they accept
    • Demand side
      • If cancellations from riders are low, they feel like the wait time is adequate given the price they have to pay
  • Repeat users
    • Supply side
      • If drivers are staying on the platform and using it on a regular basis, they are presumably happy with the service
      • In addition, the longer the tenure of drivers, the more efficient the platform becomes because experienced drivers on the platform are more likely to deliver a better experience for riders
    • Demand side
      • If riders are staying on the platform and using it regularly, then that is a success indicator as well
  • Unit economics for Uber
    • If Uber is actually making a profit from each ride facilitated, and both the driver and rider feel that a fair price was determined by the algo, then that’s an indication that the Uber Marketplace algo is doing its job correctly
    • In addition, assuming unit economics are positive for UberX, if there are enough rides to cover the fixed costs of UberX then the Marketplace is essentially a success for UberX

Being a marketplace, there are two users that Uber wants to satisfy: the rider and the driver. I think the ultimate success metric is related to how well riders and drivers get matched.

The ultimate goal for the rider is to get to their destination as efficiently as possible. Some concerns for the rider may be

• Matching with drivers
• Minimizing waiting time waiting to be matched and waiting for the driver to arrive
• Ease of pick-up and drop-off (ie. Driver coming from the opposite side of the street and the rider must walk across to meet them)
• Getting to the destination as quickly as possible

The ultimate goal for the driver is to maximize their own revenue.

Drivers achieve this through:

• Matching with riders
• Minimizing driving when they are not being paid for it (ie. driving to a rider)
• Maximizing number of rides (especially if there are driver promotions)
• Maximizing miles and time spent driving riders
• Minimizing overall gas costs

Looking through each of the drivers’ and riders’ goals, the common goal for both drivers and users is to get matched while specifically minimizing time. Minimizing time achieves multiple goals for the driver and rider as previously mentioned.

A good metric might be “% of rides matched within 2 minutes” since the current UberX experience tries to match a driver and a rider within 2 minutes. (There is a new loading screen that features a countdown in the app). We would aim to have this at 100%.

However, the trade-off with prioritizing how quickly a ride is matched might result in matches where riders and drivers are farther away from each other. If that were the case, even though the waiting time to get matched is short, the waiting time for the driver to get to the rider is long and the driver could spend time driving without being paid and incurring gas costs.

Thus, we could make the metric even more specific such as “% of rides matched within 2 minutes where time to pick-up is less than 5 minutes”.