This is a rundown of a case study I did in 2023 for a Strategy and Operation Associate/Sr. Associate role at Doordash. For those who prefer skipping the context and rationale, feel free to jump straight to the Case Prompt and Output section.

I’m documenting this to give credit to the effort I invested in this case during the recruitment phase. While it’s not an interview guide since it is based on my anecdote, it did land me a spot in the next round. I hope it serves as a helpful reference for you.

Enough with the background — let’s dive in :)

Case Prompt

Attached is a month’s worth of sample delivery data for DoorDash from one of our earliest months of operation. This isn’t the complete set, but it’s comprehensive enough where you should be able to get a good look at the business.

* Your task is to analyze this data and provide us with a set of recommendations on how to improve our business.

* Your recommendation should be focused on a specific aspect of our business (e.g. delivery performance, consumer demand, etc.) and cover the problem statement, recommendations with supporting data and assumptions, the business impact and risks

* Your deliverable can be in a slide deck (<10 slides) or word document (<5 pages), but should be “client ready” or a product you’d share externally.

the overview of the CSV file

Glossary

Below is a glossary of definitions for the variables included in the data set.

* Customer placed order datetime: Time that customer placed the order; the format is <day> <hour>:<minute>:<second>

* Placed order with restaurant datetime: Time that restaurant received order; the format is <day> <hour>:<minute>:<second>

* Driver at restaurant datetime: Time that driver arrives at restaurant; the format is <day> <hour>:<minute>:<second>

* Delivered to consumer datetime: Time that driver delivered to customer; the format is <day> <hour>:<minute>:<second>

* Driver ID: Unique identifier of driver

* Restaurant ID: Unique identifier of restaurant

* Consumer ID: Unique identifier of customer

* Delivery Region: City where restaurant is located

* Is ASAP: Equals TRUE for on-demand orders; FALSE for scheduled deliveries (e.g., a customer places an order at 10am for 12noon)

* Order total: Amount customer spent (including delivery fee); units are in dollars

* Amount discount: Amount of discounts redeemed (e.g., for referrals); units are in dollars

* Amount of tip: Amount of tip given; units are in dollars

* Refunded amount: Amount refunded to customer; units are in dollars

* Times: Time is in UTC and we operate on PT

My Approach

As someone from a non-technical background and a previous Excel monkey, I did all my analysis and visualization in Excel, paired with PowerPoint with my insights.

Upon an overview of the variables in the dataset, I broadly categorized them into 3 buckets:

  1. Logistics: This involves all timestamp-format data, where I can see how operational efficiency can come into play
  2. Revenue: Data we can derive Order# and Order$ from, which is a key KPI that indicates performance related to regions/time of the day
  3. Services: Variables like Tips/Refund/etc that hint service satisfaction, which can be an indirect lever to drive revenue

I think the best way was to pick one category for this exercise. While diving into logistics would require additional timestamp data cleaning, and determining the correlation between service satisfaction and revenue-related variables demanded a more quantitative approach. Based on the following considerations:

a. I aimed for a concise output, favoring clear insights and straightforward visuals over excessive quantification.

b. In the context of the interview, I didn’t believe that investing extra time in rigorous quantitative analysis would significantly set me apart

Therefore, revenue seemed to be the best option for this exercise. My slides are attached in the subsequent section. I know that’s probably one of the most vanilla slides you’ve seen in your life but I am working on it :(

Output

Overview:

This analysis focuses on the monthly order trends for the current period, breaking down by Region, Order Type, Restaurants and Customers, aiming to spot any patterns within the order distributions across these segments.

The goal here is to gain insights into the revenue composition for the current period. This information guides our priorities, helping us maintain our current position, develop and scale our business, and ensure long-term sustainability.

Last Words

Of the several data-driven case studies I did in 2023, this Doordash case stood out as one of the most friendly (don’t confuse that with easy) ones. There was a clear business model and revenue proposition. The dataset was relatively clean and simple while giving me enough autonomy to pick an angle in their business to dive into. Obviously the company was keen on candidates that were analytical and logical, while also creative. It instantly brought to mind another interview I had, where an overwhelmingly messy dataset was dumped on my lap, and it took me over 1 hour to comprehend+clean the data before diving into any analysis, got me questioning that company’s data collection process or whether I had inadvertently applied for a data wizard role.

Also in case you are wondering, my recruiting timeline for Doordash looked like this:

Apr 25, 2023 — HR phone screening

Apr 28 — Received this take-home assignment (to be submitted within 48 hours)

May 3 — Informed that I passed the case and asked about my availability for the next round

May 10 — Two back-to-back 1-on-1 interviews with two different managers from the team, which consisted of both case and behavioral questions

May 11 — Got the “You weren’t selected” e-mail

Thanks for making it this far. Would love to hear your thoughts! :)