The person creating this sheet would need to capture the underlying logic of their business through the formulae in the sheet, and then copy-paste the generated values into a different sheet which would be then shared with the candidates. Those values would be static, and unlike this sheet, would not be regenerated every time this sheet reloads.

Since that data sheet would not have the formulae, one of the goals would be to see if the candidates can reverse engineer that by looking at the data itself, and understand and present the business logic as part of their assignment submission.

A good versus great response for this exercise would be if the candidate could parse out how the business works.

For example, for improving revenue, could they figure out what the refund margin is, and suggest reducing it? Or suggest a flat refund amount, or future credit and so on.

The prompts / hints given to the candidates were fairly open ended, and used to set a general high level direction, and it was up to the candidate to decide where they wanted to take the analysis themselves.

We deliberately did not dictate a presentation format such as slides, spreadsheet, a doc, a PDF or anything else.

PMs use all of those formats to pitch ideas to others, so we left this open-ended to give the candidate the most flexibility to choose the format that they were most used to and felt most comfortable in presenting.

We were looking for the responses to be well thought out with a couple of deep insights, good visualizations, and prepared as if it was a readout to be shared with others.

In terms of grading, presentation and visualization was weighted at 30% and used as a filtering criteria - if the candidate did an outstanding job, they would get 30%, but would get dinged if they did a bad job. And the insights and ideas were 70% of the grade.

Compared to the companies that use an in-person presentation for take home exercises, we chose not to do that for a few reasons:

1. We were evaluating for verbal communication in other impromptu case questions that we were asking the candidates as part of the other interviews anyway.
2. Unfortunately, most people dropped out when they received this exercise, which is why we removed it later on. But of those that did submit, many of them were good enough to the point where we didn't feel the need to talk to them in person.
3. The exercise was to evaluate the candidates written skills. And the response format was intentionally required to be a readout to be shared asynchronously vs in a synchronous meeting. That easily helped separate the good from the great - the best readouts had very clear takeaways and did not need a further in-person discussion. The ability to communicate clearly through written communication to avoid extra in-person meetings was also an attribute that we were also looking for through this exercise.

Vishal Kapoor

This dataset generator is similar to one I used for product analytics take-home exercises. Use the "Open In" button to access your own copy of the sheet.

Dataset Generator for a Delivery Company Analytics Case

You would typically use such an analytics take home exercise generator sheet If you're in a technical / backend heavy team (like revenue or forecasting), but if you’re hiring for a user or front end facing team, then you may not need this.

Dataset Generator for a Delivery Company Analytics Case

Dataset Generator for a Delivery Company Analytics Case

Data sheet

Candidate Exercise Prompts

Sign up to see more