How to pass the Product Analytics interview in tech

I had 8 final round interviews for Product Analytics roles and received 6 offers. I also worked in product analytics in tech for 5 years. This is my advice on how to pass the interview

Sep 09, 2024

What is Analytics vs. Data Science?

There are many articles that have discussed the distinction between analytics and data science roles already. It is confusing because a lot of analytics roles are actually titled as data scientist. Here is my opinion of the distinction between analytics vs. data science, but I also link some articles and forums that discuss the distinction.

In my opinion, analytics roles are focused on using descriptive statistics to determine what happened and why. Analytics roles should also be able to design and execute AB tests, know a bit about causal inference, and of course, be able to generate reports and dashboards. There is little to no focus on building models to put into productions, and there is little focus on prediction. Data scientist or machine learning roles on the other hand, do put models in production and are more focused on making good predictions.

Some articles that discuss the difference between analytics and data science:

Northeastern University article

Coursera article

A reddit discussion

Analytics job titles look like:

Product Analyst
Data Scientist, Analytics
Product Data Scientist
Business Intelligence Engineer (this is specific to Amazon)

How to pass a Product Analytics Interview (assumes that you can get the interview)

Unfortunately, I don’t have much advice on how to get an interview. What I’ll say is to emphasize the following skills on your resume:

SQL
AB testing
Using data to influence decisions
Building dashboards/reports

And de-emphasize model building. Not that model building is not valued, but it’s not the skill that is sought after in analytics roles.

Assuming you get the interview, here is my advice on how to pass it:

You have to be able to pass the SQL screen. For this, I used two resources: Stratascratch and Interviewquery. You only need to subscribe to one as they both provide pretty similar resources. This assumes that you already know the SQL basics and just need practice solving SQL problems under pressure. If you don’t know any SQL then google some free SQL resources.
- I personally preferred Stratascratch. I felt like the content was better for folks looking for analytics roles, while Interviewquery was a bit more geared towards Machine Learning roles. But honestly either one is fine.
If you pass the SQL screen, you will likely be asked some case study type questions. To pass this, you’ll like need to know AB testing, as well as getting used to using analytics to think through and solve problems. For this, I recommend two resources:
- Trustworthy Online Controlled Experiments by Ron Kohavi - This book is the gold standard for learning about AB tests. It is a very readable book and does not require a deep math or statistics background.
- Ace the Data Science Interview by Nick Singh - No need to read the whole book. There are two chapters in here that are focused on analytics interviews: The Product Sense and the Case Studies chapters.
  - Alternatively, this article by Interviewquery provides a lot of case question examples, although it doesn’t provide sample answers. All of them relevant for analytics role case interviews except the Modeling and Machine Learning section.

Disclaimer: Some of the links are affiliate links if you’d like to support me. If a resource that I recommend costs money, I have purchased and used it to land product analytics jobs, without any compensation. I literally bought those services or products based on my own online research, and am now recommending it because I found it useful. If you think the resource might be helpful but you don’t want to use the affiliate link, just google the name.

What type of work do Analytics folks do?

Here are some example projects that I have worked on, or that I have seen my friends work on.

Your team has identified 10 potential projects to work on in the next quarter. They want to know how to prioritize it, so they ask you to estimate the ROI of each project so that they can use it to inform prioritization of the projects
You work for a gaming company that just released a new game for mobile. They ran a 1 month long advertising campaign across the entire US. Because it was a nationwide marketing campaign, there was no control group so it was not AB tested. You are asked to determine the impact on app downloads that can be attributed to the campaign
You helped your team to run an AB test on a feature to increase conversion rate on the website (think changing the “buy” button from red to green). This AB test was 2 weeks long. You can easily calculate the impact over that 2 week period, but the leadership team wants to know what the impact would be of implementing this change over the next 12 months.
You work for Uber Eats. There was a system issue that caused every customer in LA who ordered between 5pm-5:05pm to have their order canceled. You are tasked with determining what amount to refund those customers for their troubles, such that Uber has the highest chance of retaining those customers, but also not so much money that Uber is just losing margin.
The marketing team wants to segment their customers into groups such that they can create customized marketing messages for each group. You are tasked with this customer segmentation task
You work for Netflix. They want to know the leading factors of churn so that they can take measures to prevent customers from churning.
Your company sends a lot of marketing messages to customers, including email and push notifications. Sometimes, customers will unsubscribe. Your company wants to know how much revenue they lose when a customer unsubscribes
Measuring profit margin in AB testing is too noisy. Identify some leading indicator metrics that we can measure in AB testing that are less noisy and quicker to measure than profit margin, but that provide good signal on how profit margin is impacted.

What technical skills are required for Product Analytics roles?

In my experience, SQL is the top skill. Every single analytics role that I interviewed for tested my SQL skills. Only 2 tested Python or R. You do need to know enough Python or R to do visualizations, data munging, and running regressions. Additionally, you need to know enough math statistics to understand AB testing and common causal inference techniques. This should be undergraduate level.