Framework for Promo Prompts
DoorDash (and a few other tech companies) that are data-driven like to ask questions about how you would set up a pricing promotion. They want to see how you think about tackling data problems and experiment setup.
DoorDash likes to ask about an off-line promotion. This throws off people who are used to online data at their finger tips.
Common problems I have seen from candidates:
Using online solutions anyway
Trying to create forced situations in physical spaces
Ignoring the elephant in the room and not discussion tradeoffs
Only discussing controlling experiment by running in different stores
To keep yourself centered during the interview, it is a good idea to have a basic framework. I like this five-point framework.
Clarify
Deteremine Decision Metric
Set-Up Experiment
Test & Learn
Discuss Other Measures
Clarify
Make sure you understand the goal and complications. For example, is it top-line revenue or longer-term customer loyalty. What are short-term and longer-term goals.
Determine Decision Metric
Don’t jump into the experiment setup without knowing what it is you are measuring. It could be revenue, but it might not be. Knowing how you are going to measure success shows you can prioritize your decision.
Are you looking to test your ability to bring them into the store? Or do you hope to increase their spend while in the store? These are very different goals with very different experiment setups.
Set-Up Experiment
This can be really tricky for offline (think the Target promo question). You can’t control entry, exit, and exposure in the same way in a store as online. People will see displays, and you can’t be sure how much that impacted purchasing decisions, maybe they came in the store with that intent? How do you account for test and control groups?
Test & Learn
What are you hoping to learn? How do you decide the limits, or lack thereof for your experiment? How can you modify your experiment based on poking questions from your interviewer?
Other Measurements
You had to determine your decision metrics. What are creative ways of measuring your desired goal and unintended consequences? How do you isolate for noise? You will need to look at secondary metrics for both negative impacts to avoid and positive learnings that could inspire new experimentation paths.
Tradeoffs to Consider
If my goal is driving top-line revenue I have a different goal than if I want to move merchandise to reduce storage costs.
My decision metric might be increased basket size determined by items in the basket or total revenue.
Do I count all receipts in Store 1 vs Store 2? Or do I consider all receipts without the sale items vs receipts with the sale item in the same store?
Do I consider discounts across the board or buy-one-get-one promos?
Other Resources
For more on DoorDash interviews, check out this article on prioritization prompts and here for a list of practice prompts for all types of DoorDash interviews.
