Skip to content

How to Ace Your PM Data Interviews

A brief guide for winning your analytical job interviews
Sessions are down 20% today - why?
(Want to skip ahead to a sample question? Start here

- currently working on Product, Data and Growth at Coda, previously startup Founder/CEO, and before that a PM at Google and YouTube. I’ve been both the interviewer and interviewee for product analytics questions countless times. This is the first time I’ve shared my playbook.
, and

You’re interviewing for your dream job as a
at Unicorn, Inc. You’ve researched the company, used the product, and even have your list of ideas ready for how to improve it. But you didn’t see these data questions coming and don’t know where to begin. Why didn’t you click on that link “How to Ace Your PM Data Interviews” you saw scrolling through your feed the other day? Just kidding! Let’s dive in. 👊

How to read this
My hope is that after reading this post you have a cheat sheet for how to answer common
analytics interview questions and you have some easy next steps to get even better prepared.

This page focuses on high level context and strategy. We’ll walk through kinds of questions you’ll be asked, each of which have deep dives on their own pages. We’ll break down what interviewers are looking for and in the following pages we’ll dig into detailed sample questions, with example strong and weak answers at each step.

The meat of this is post in the detailed walkthroughs:
, and
. There’s also an extensive bank of
and additional

While we’ll focus on
interview questions, in practice these same kinds of questions often show up for Software Engineers, Marketers, Data Analysts, and other roles.

What interviewers are looking for
Keep in mind interviews are more about understanding how you think than getting a “right answer.”
Candidates often make the mistake of getting caught up trying to get to a “right solution” and skip explaining their reasoning. Reasoning is
the main thing
an interviewer is trying to assess. So it’s much better to have amazing reasoning but not fully get to a final answer than an answer that’s hard to contextualize or reason about.
Common pitfall
💭 Structured thinking
Uses a framework to solve a problem methodically. Articulates and rejects hypotheses. States assumptions explicitly, and what could change them. Communicates goals that drive decisions.
Jumps straight into details without setting any context. Assumes something unstated and incorrect about the setup.
🤔 Asks good questions
Asks questions that bisect the solution space and give the most information on what to do next. For more on what this looks like, check out
Overly focused on the wrong details. Asks lots of small clarifying questions when clearly trying to buy time to think.
🦅 Changes altitudes
Able to go from high level abstractions to nitty gritty details. Able to talk about different user segments and how behaviors and characteristics vary across them.
Jump to an answer but miss the chain of reasoning on how you got there.
📓 Demonstrates experience
Shows expertise by connecting answers back to specific personal experiences. Demonstrates familiarity with terminology, methodology, and best practices. Note this skill is called “demonstrates experience” rather than “has experience”: it’s not enough to have relevant knowledge, you have to show it and connect it effectively.
Speaks in generalizations without giving specific examples from own experience.
💬 Communication
Good communication skills. Takes time to think when needed, signposting thinking, asks questions, isn’t defensive when challenged, reads when the interviewer is giving hints.
Feels obligated to start talking immediately and just starts talking, clearly stalling for time.
There are no rows in this table

Back when I was at Google, most interviews were scored on a 5 point system (1 being poor and 5 being strong). Here at Coda, we use a 4 point system - which is much clearer: you cannot do OK. As an interviewer, if I’m on the fence, that rounds down to a no.
1 - strong no 😡
2 - no 🙁
3 - yes 🙂
4 - strong yes 😍

Three types of data interview questions
There are three main types of data interview questions. We’ll break each down in detail in the following pages.
Dig into a change in metrics to uncover why it changed.
Decide what metric a product should optimize for and why.
Identify, compare or use growth levers for a feature or product.

Note these types of questions go by many different names at different companies. For example, Facebook has a mix of these under an Execution Interview header, Thumbtack calls these Analytical Interviews, and here at Coda we have a Marketecture Interview. The reality is these questions types aren’t as discrete as I outline here, but exist on a spectrum. That said, I’ve found these three types useful as “corners of the tent” to show a range of problem solving formats, with the more common case somewhere in the middle.

Three simple steps to frame your answers
Here’s my simple framework to use on all of these questions:
Step 1:
set context.
Reiterate and clarify goals, constraints, and important factors to the problem or solutions.
Step 2:
go broad.
Brainstorm solutions to the prompt, get more constraints from the interviewer if appropriate, and discuss pros and cons.
Step 3:
Select your solution. Describe mitigators, things that would change your answer, and followups.

My friend
@Dina Levitan
pointed out this is similar to
. Here’s what this looks like applied to each kind of question:
Question type
The arc of this question
Step 1: set context.
Clarify your understanding. Ask follow-up questions to better diagnose what is going on.
Step 2: go broad
. Through the process of digging in, articulate a range of hypotheses. Describe the key data points for verifying what is going on.
Step 3: converge
. Settle on one hypothesis and flesh it out with more detail. Propose a few solutions given the hypothesis, and select and justify one for execution. Discuss how you'd measure that your solution is working.
Step 1: set context.
Describe the highest level mission and business goal of the product. You'll derive your goal setting metrics from this. Discuss your different user types and which are core vs growth areas.
Step 2: go broad.
Propose a few candidate high level metrics that could illustrate progress towards the mission. Discuss flaws and pros/cons with each metric, and ways to mitigate the downsides.
Step 3: converge.
Select the final choice of key metric, and describe a process and secondary metrics that you could use to prevent "overfitting" to the metric.
Step 1: set context.
State your product mission, business goal, and highest level metric. Describe who your product's different user types are and why they use the product today.
Step 2: go broad.
Return to the growth question and propose a few different approaches, and discuss pros and cons.
Step 3: converge.
Select one of your approaches. Describe mitigating factors and learnings or new information that could change your approach.
There are no rows in this table

It’s worth calling out that having a good framework for your answer is necessary but not enough by itself. There’s a common type of poor interviewee who is overly focused on a high level framework, and fails to engage in the gritty detailed discussion and trade-offs of concrete options and ideas. Don’t be “just a framework person” - make sure you continue to connect it back to the details.

If it sounds too simple and easy, you’re half right. It is simple, but it’s not easy. Like most things, it comes down to how you do it. The best way to do it well is to practice. 👇

Practice makes perfect
Interviewing is a skill that, like all skills, you get better at the more you do it. One of the best ways to practice is to go through sample questions with friends or colleagues and have them give you feedback. Getting more examples under your belt always helps. In that spirit, we’ll walk through example questions together and analyze a range of different sample answers. Or you can skip ahead to the
. Read on to the next pages for more detailed tear-downs per question type:

👉 Next, let’s jump into
Want to print your doc?
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
) instead.