This was a case I did in Aug-Sept of 2023, which was the initial screening for a Strategy & Operation role at Clipboard Health. Again like my other case study posts, this isn’t intended to serve as an interview guide. This submission got me into a subsequent 30-min interview where I walked through my thought process with a member of the team. While it may not be an exhaustive guide, this post can still serve as a useful point of reference and offer useful insights. Without further ado, let’s dive in.

Case Prompt — Lyft Toledo

Clipboard’s marketplace users “book a transaction in the future,” as is true for other apps with which you may be familiar. So we use the analogy below to leverage that familiarity.

Pretend you’re the pricing product manager for Lyft’s ride-scheduling feature, and you’re launching a new city like Toledo, Ohio.

The prevailing rate that people are used to paying for rides from the airport to downtown (either direction, one way) is $25. The prevailing wage that drivers are used to earning for this trip is $19.

You launch with exactly this price: $25 per ride charged to the rider, $19 per ride paid to the driver. It turns out only 60 or so of every 100 rides requested are finding a driver at this price.

(While there is more than one route to think about in Toledo, for the sake of this exercise you can focus on this one route.)

Here’s your current unit economics for each side:

  • Drivers:
  • Customer acquisition cost (CAC) of a new driver is between $400 — $600. CAC is sensitive to the rate of acquisition since channels are only so deep.
  • At the prevailing wage, drivers have a 5% monthly churn rate and complete 100 rides / month
  • Riders:
  • CAC of a new rider is $10 to $20 (similar to driver CAC it’s sensitive to the rate of acquisition, since existing marketing channels are only so deep)
  • Each rider requests 1 ride / month on average
  • Churn is interesting: riders who don’t experience a “failed to find driver” event churn at 10% monthly, but riders who experience one or more “failed to find driver” events churn at 33% monthly

You’ve run one pricing experiment so far: when you reduced Lyft’s take from $6/ride to $3/ride across the board for a few weeks, match rates rose nearly instantly from 60% to roughly 93%.

Your task is to maximize the company’s net revenue (the difference between the amount riders pay and the amount Lyft pays out to drivers) for this route in Toledo for the next 12 months. Let’s assume that you cannot charge riders more than the prevailing rate.

The core question is: how much more or less do you pay drivers per trip (by changing Lyft’s take)? Your goal is to maximize net revenue for the next 12 months on this route.

As you tackle this case study, keep in mind that Clipboard’s Product Team dives deep into the numbers and has a bias toward action. And we have real fun doing it!

Output for the Case Study

The ideal output from this exercise is a written document backed by analysis, which is the type of output our team members produce regularly. As mentioned in our blog, we do not think longer cases are “better” and value clarity far more than length.

My approach

The case presented a typical 3-sided marketplace scenario, which is applicable to business models for companies like Uber, Doordash and Lyft, etc. For those unfamiliar, Clipboard Health operates on a similar model, serving as a matchmaking platform connecting nurses with healthcare facilities for shift work

The primary objective of this case is to determine the optimized revenue, with the main variable we can directly manipulate being Lyft’s take on each ride. Additionally, it’s crucial to consider the following factors:

  1. Multiple Influencing Variables: Numerous variables can impact the output, underscoring the importance of making reasonable assumptions to narrow down our levers. (Assumption is highly important and what the interviewer cared the most)
  2. Timeline Consideration: The analysis is focused on the next 12 months. While many discussions revolve around “long-term” business growth, in this case, the emphasis lies on factors affecting the immediate future.

For my submission in the upcoming section, I built a P&L model in Excel and consolidated the work in a Word doc. Visualizations were done in Excel and the “goal seek” function was used to find the optimal point.

Output

Case Background and Objective:

As the Project Manager overseeing the ride-scheduling feature at Lyft, our primary objective is to optimize Lyft’s revenue over the next 12 months in preparation for the launch of our services in Toledo, Ohio.

This case study seeks to dissect the underlying problem, with a particular focus on narrowing down the key variables to driver compensation. Within reasonable assumptions, we will develop a model to explore the singular influence of Lyft’s revenue per ride on its 12-month revenue. Subsequently, we aim to determine if an optimal Lyft’s revenue per ride exists to maximize the 12-month revenue.

Key Variables and Levers:

Drivers

  • Prevailing rate of $19, at which:
  1. Match rate = 60%
  2. Completes 100 rides/month
  • Monthly churn rate: 5%
  • Customer Acquisition Cost (CAC): $400–600

Riders

  • Prevailing rate of $25 (not subject to increase)
  • Average 1 ride/month
  • CAC = $10–20
  • Churn rates:
  1. 10% if no “failed ride request” is experienced
  2. 33% if “failed ride request” is experienced

Lever under Lyft’s control:

  • Lyft’s share per ride, which directly affects driver’s pay
  1. Currently at $6
  2. $3 under test scenario

For additional assumptions, please refer to the appendix at the end of this report.

Process of Analysis:

1. Dissecting Lyft’s Revenue

We start by breaking down Lyft’s monthly revenue, which can be expressed as the product of the number of monthly completed rides and Lyft’s revenue from each ride. Mathematically, this can be defined as:

Lyft’s monthly revenue = monthly completed rides*Lyft’s revenue from each ride

= monthly completed rides*(rider’s pay — driver’s pay)

= (monthly demand*match rate)*(rider’s pay — driver’s pay)

Our current analysis reveals a mismatch between demand and supply, with demand surpassing supply, as evidenced by the match rate being less than 100%. Assuming that churn occurs only on a month-over-month basis and monthly demand remains constant within a given month, the focus shifts to finding the optimal driver’s pay that balances monthly completed rides and Lyft’s revenue per ride mix to maximize 12-month revenue.

2. Impacts of Driver’s Pay on Match Rate & Churn Rate

To examine how driver’s pay affects completed rides, we can analyze the impact on match rate and churn rate, following the logic below:

Drivers pay increase-> higher demand -> higher match rate -> lower average rider churn rate

For simplicity, we assume linear correlations. (However in real practice the relationship should be tested from data) Based on linear regression and available data, the following correlations can be plotted:

Notes:

- Under the prevailing pay of $19, drivers won’t accept any rides, i.e., match rate reaches 0 when driver’s pay falls below $19

  • Match rate reaches 100% when driver’s pay increases to $22.64 (derived from linear relationship), above which ride supply will be over demand

Notes:

  • When driver’s pay reaches $22.64, match rate equals 100%, meaning all ride requests from riders will be fulfilled, leading to a churn rate of 10%

3. Finding the Optimal Driver Pay to Achieve Maximum 12-Month Revenue

Considering the impact of driver’s pay on both churn rate and match rate, we can derive its influence on demand and, consequently, 12-month revenue — the ultimate objective. We proceed with the following steps:

a) For simplicity, assuming there are 1000 active users (rider) of Lyft from month#1 (Under the same linear relationship assumption, the optimal Driver’s Pay derived from this model would also be applicable for the real active user data).

b) Based on our previous analysis, applicable driver’s pay falls within the range of $19 to $22.64:

i. When driver’s pay < $19, drivers won’t accept any rides

ii. When driver’s pay >$22.64, ride supply is over demand which is inefficient

c) We select 10 values within the ($19-$22.64) range at equal intervals as the variables.

Using the data, we construct the following model:

The resulting chart displays the relationship between driver’s pay and 12-month revenue:

With the Excel solver function, we identify the optimal driver’s pay as $21.71 (Lyft taking $3.29 per ride). At this point, we achieve the maximum 12-month revenue and the highest Lifetime Value to Customer Acquisition Cost (LTV/CAC) ratio of 1.6X.

Appendix:

1. Assumptions:

a. Churn rate = (lost customers/total customers)*100

b. Lifetime = 1/(Churn Rate)

c. Monthly Average per rider revenue = Lyft’s revenue per ride*Match Rate*Monthly average ride demand

d. LTV = Lifetime(months)*Monthly Average per rider revenue

= Monthly Average per rider revenue/Monthly Churn Rate

2. Rider Churn Calculation:

Current situation — Lyft takes $6 per ride and driver gets paid $19

  • Ride request fulfilled = 60%
  • Ride request unfulfilled = 40%
  • Average Rider Churn = 60%*10%+40%*33% = 19.2%

Test situation — Lyft takes $3 per ride and driver gets paid $22

  • Ride request fulfilled = 93%
  • Ride request unfulfilled = 7%
  • Average Rider Churn = 93%*10%+7%*33% = 11.61%

Additional Context and Thoughts

In my opinion, after making a proper and relevant assumption the math and analysis in the case becomes pretty straightforward. The part I spent most of my time on was figuring out how to construct and present this case clearly and concisely. It’s worth noting that Clipboard Health has a writing-heavy culture (likely due to team members being across different time zones), which I tried to cater to. As an ESL, constructing a comprehensive passage within a soft time constraint was a bit challenging.

In the subsequent interview, the pivotal question revolved around identifying the most crucial assumption in my case and explaining its significance, and I was mainly challenged on my rationale for the linear correlations assumption. The follow-up questions dug pretty deep into how I would approach assumption testing, which covered data collection, testing groups, etc.

From my experience, I would say the interview was quite unique, interesting and thought-provoking. When I was prepping, I thought the case was straightforward forward so I didn’t expect the interview to get to that level of detail and depth. The Conversation I had with the team member was engaging and fun, and really gave a glimpse of how the product team at Clipboard Health works — following the first principle comprehensively.

A quick heads-up — The company has a not-so-favorable reputation on several job boards and communities for its recruiting process:

They’re notorious on keeping the S&O and product manager roles open with no real timeline for hire

The case was long and feedback is notably absent after rejection

Despite the bad reviews, for me personally, I still think the case was a great practice that helped me stay fresh.

Anyway, hope this helps and let me know your thoughts!