Data analytics interview questions sometimes operate like a flowchart. They can be woven into other interview questions about a broader problem.

For example, often in product manager interviews, an analytics interview question can be intertwined with the original question.

If a PM was asked to design an application to show their technical spec and product management skills, they might then be asked what metrics they would track to prove the success of their application down the line to test their analytical skills.

Most data analytics problems are framed around a case study format where we’re given a problem statement and have to figure out a way to solve it. However, because most data analytics case studies cover a wide breadth of skills, they usually can be separated into different parts.

Data Analytics Case Study

As mentioned before, data analytics case studies encompass an analytics problem statement and potentially ask you to write code or a query to return some metrics or analyze the data to prove a hypothesis in real-time.

Example Scenario:

   1. We’re seeing a 10% decrease in user signups over the past week. How would you investigate what’s happening?

Additional Questions:

1. Given your hypothesis for what’s happening, write a query to see if your hypothesis is correct.

2. Suppose you’re working at Reddit - how do you determine if a subreddit is performing well?

3. Let’s say you’re given all the different marketing channels and their respective marketing costs. What metrics would you use to determine the value of each marketing channel?

Defining Business Metrics

This is a business. Here’s a problem. How would you develop metrics to see the business’s success or the problem?

Example Scenarios:

  1. Suppose you’re working at Reddit. How do you determine if a subreddit is performing well?      What metrics would you track?

  2. Here is a table schema of three tables: posts, users, and comments. What metrics would you      track to determine whether a user is a “healthy” forum member?

  3. Let’s say that we run an eCommerce business. Create a dashboard that could monitor the      health of this business.

Technical Data Analysis

The technical data analysis section involves taking the context from the problem and analyzing the data or writing a query to dive into the data. Mainly this tests SQL and technical data skills in Python or Pandas/R. These are commonly added questions on top of the data analytics case study.

Example Scenario:

  1. Let’s say you come up with a metric that is the “Percentage of daily active commenters”. Now      the interviewer will ask you to write a query to pull that metric by day.

  2. Analyze the dataset we give you and present insights from what you found.

Here, the data is real and not just a schema. It could likely be real data from the business. Sometimes the interviewer will have this planned out, usually by an existing analysis by a data scientist on the team that turned into a strong report, and now they’re using it to evaluate candidates.

Other times they just took an excerpt of their dataset and looked to see if candidates could find any kind of insight for a problem they were struggling with while working together on it.

Data analytics is tested in different ways, and there isn’t a specific grading system for how it all occurs, but it depends.

Meta (Facebook), for example, asks a data analytics question in every one of their interviews that follows the format of first, giving the users a set of tables, showing how they can run code in SQL or pandas to look into the tables, and then asking them to provide a metric for a specific situation.

For example, given a table of users and friend requests, come up with a metric that demonstrates at what point users stop sending as many friend requests. Once you develop a metric, you’ll be asked to write the query to get the specific information.

Facebook’s interview format is an excellent example of providing structure to data analytics. While we defined data analytics above as a problem with an input of a dataset and a hypothesis, and an output metric, this kind of interview problem assesses both the ability to hypothesize and create a suitable metric while also then being able to have the technical knowledge to retrieve such a metric.