Meta Product Growth Analyst (L4/L5) Interview Experience
Meta · Mid level · Product Analyst
At Meta, a product growth analyst is not considered any less than a PM. About 40% of them convert into PMs in one or two years because you are expected to know how even a red button versus a blue button makes a change.
Interview process
The Meta Product Growth Analyst loop started with a recruiter screen, then a 45-minute case study plus SQL screen, and then a final loop with two product improvement rounds, one SQL round, one execution round, and one problem-solving round. The whole process is very product-growth heavy, which means they expected me to think like a PM who can also talk metrics, experiments, and data.
The first screen was more guided, but the final product improvement rounds were very open-ended. I think this is where most people struggle: they jump to solutions without defining the metric, the funnel, or the user segments. SQL mattered, but the bar was clearly lower than for a data scientist. I ended up joining, and my biggest takeaway was that the real signal came from whether I could stay structured, generate a lot of ideas fast, and still prioritize, as I would actually ship something in a big company.
Interview rounds · 7
- 1
Recruiter screen
BehavioralI started with a pretty standard recruiter chat where they mostly wanted my background and whether I actually fit this product, growth, and data mix.
Q1. What have you been doing, and can you walk me through your background?
How they answeredI kept this high level and focused on the growth side of my work, how I think about user behavior, metrics, and experimentation, and why I fit a role that sits between product and data science.
- 2
Phone screen
AnalyticalProduct StrategySQLData AnalysisThe first real screen was a 45-minute case study plus SQL round, and it was more guided than the final loop because they gave me the problem and pushed me to show structure.
Q1. Comments on group posts are decreasing day by day. What data will you pull to look at why it is happening?
How they answeredI would start by defining the user journey and the metric before touching solutions. I would check whether group membership is down, whether fewer posts are being created, whether viewers are seeing posts but not engaging, and whether the drop is concentrated in certain post types or user cohorts. I would segment hard, like text vs photo vs video posts, and whether people are viewing, liking, commenting, or just ignoring. The whole point is to show multiple hypotheses instead of jumping straight to ideas.
Follow-up questions- Are my users going down, or is it just that the number of groups is less?
- Are people leaving groups, or is the issue within existing members?
- What cohort is actually affected?
Q2. What are you going to do about it? How are you going to increase it?
How they answeredI would first pick a very clear metric, like commenters over viewers or commenters over people who already engaged, and explain why that cohort is the low-hanging fruit and will get statistically significant results faster. Then I would prioritize ideas for the people already in the funnel instead of saying something vague like 'get more people into groups.' I would segment which post formats are getting ignored and then propose things like GIFs, automated responses, or other prompts that actually make commenting easier. They really care that I prioritize one path and explain why it is executable.
Follow-up questions- What exact metric are you trying to improve?
- Why are you choosing that metric over another one?
Q3. Can you pull the percentage of users posting Reels within the United States?
How they answeredI would expect something simple here, basically a filter and join kind of question, not some rank-heavy SQL. The bar for product growth analyst SQL felt more about whether I can get the right answer quickly than whether I can write the fanciest query.
- 3
Final / onsite round
Product StrategyProduct DesignAnalyticalExecutionThe hardest part of the loop was the open-ended product improvement round, where they gave me one line and expected me to drive the whole thing without getting lost in the dense forest.
Q1. How would you increase the number of shares a Reel has?
How they answeredI would break the funnel down all the way from visibility to views to repeat views to likes, comments, and shares, then define exactly what I want to move. I would say whether I care about shares over viewers, shares over engagers, or something else, and why that cohort is the best low-hanging fruit. After that I would generate a big list of ideas, around 10 to 15, then bucket them into on-platform and off-platform ideas and prioritize what is fast, executable, and likely to show measurable lift. I would also keep legal, marketing, and consistency constraints in mind because not everything can actually ship.
Follow-up questions- What is the ideal user journey and the non-ideal user journey here?
- What metric are you actually improving?
- Why did you choose that numerator and denominator?
- Who is the easiest set of users to target first?
- Which ideas would you prioritize, and why?
- 4
Final / onsite round
Product StrategyExecutionAnalyticalThe second product improvement round usually had an ads angle, and it tested whether I could scope the funnel correctly instead of solving the wrong problem.
Q1. We want more people to click the Boost button on business pages. How will you do it?
How they answeredI would first clarify whether the bottleneck is top of funnel, like not enough business pages, or mid funnel, where existing business pages see the button but do not click it. Then I would define the user journey, pick the metric, and spend the rest of the round on creative ideas and prioritization. The key is not to wander. I would get to metric and data in the first 5 to 10 minutes, then spend most of my time on the idea set and why one idea is more executable than another.
Follow-up questions- Should we convert more pages into business pages first, or do we already have enough business pages and the problem is that they are not clicking Boost?
- What data and metric would you use?
- Which ideas would you choose first?
- 5
Final / onsite round
SQLData AnalysisAnalyticalThe dedicated SQL round was still not data-scientist-level hard, but it moved fast, usually with multiple questions and a metric follow-up if I finished early.
Q1. Can you pull the count of users in the US who use Instagram and Threads both?
How they answeredI would solve the SQL quickly with straightforward joins and filters, then switch into diagnosis mode. If the metric is going low, I would ask whether the numerator is dropping or the denominator is changing, then segment by user type, platform, and other obvious cuts to isolate where the decline is coming from.
Follow-up questions- Suppose this metric is going low. How would you identify why?
- 6
Final / onsite round
BehavioralExecutionProject ManagementCross-FunctionalThe execution round was different from product improvement because it mixed behavioral questions with prioritization and sizing, so it was less about raw ideation and more about making a call under constraints.
Q1. Tell me about a time you had a disagreement or had to deal with prioritization.
How they answeredI approached this by showing how I prioritize ideas based on impact, feasibility, and timing, especially whether something can actually ship in time for the quarter. I would also speak to how I handle pushback from partners and how I narrow choices when there are legal, marketing, or engineering blockers.
Follow-up questions- How did you decide what to do first?
Q2. You have bandwidth for only one thing. Would you implement automated GIF replies or add a blinking comment button, and why?
How they answeredI would size both. I would look at expected lift, target cohort size, engineering effort, and whether we already have evidence from a similar A/B test somewhere else. The point is not just to pick something clever. The point is to justify why one option is the better bet right now.
Q3. Suppose you have zero marketing dollars. Give me at least six different ideas to increase the number of users on Threads.
How they answeredI treated this like a creativity round but kept it structured. I would generate a spread of on-platform and off-platform ideas, tie them back to user psychology, and use things like motivation and social proof instead of paid acquisition. They really want volume of ideas here, but they still want them categorized and prioritized, not random.
- 7
Final / onsite round
AnalyticalData AnalysisExecutionThe problem solving round felt more like metric debugging, where they wanted a clean diagnostic tree and strong segmentation before any conclusions.
Q1. Instagram daily users are increasing, but Messenger daily users are decreasing. What will you do, and why do you think that is happening?
How they answeredI would start with sanity checks before making up a product story, like whether this is a dashboard issue, a logging problem, or just a denominator problem. Then I would segment the decline by platform, age, and Messenger type to see where it is actually happening. I would be very specific about the metric because sometimes the numerator is falling and sometimes the denominator is changing. This round is really about asking the right questions in the right order.
Follow-up questions- Is this a false alarm or a data logging issue?
- Is this a sudden change or a longer trend year over year?
- Which Messenger segments would you cut first?
- Are you segmenting by platform like Android vs iOS, or by age?
Tips from the candidate
If I were helping a friend prep for this, I would say do not jump into solutions in the first two minutes. First define the metric, the user journey, and the cohort you want to move, then segment the problem properly. For product improvement, I would practice coming up with 10 to 15 ideas quickly, but not as a random list. Bucket them, like on-platform vs off-platform, and then explain why one is the low-hanging fruit and why it is executable. I would also prep basic SQL for speed, not fancy tricks, and I would absolutely study user psychology because a lot of the strongest ideas come from motivation, friction, and social proof.
Company culture
What I felt at Meta was that product growth analysts are not treated like narrow analysts. They are expected to wear multiple hats and in a lot of ways they are evaluated almost like PMs who can speak data. The SQL bar is lighter than for data scientists, but the creativity and prioritization bar is much higher, especially in the final product improvement rounds. Interviewers care a lot about speed and structure because these questions can turn into a dense forest if I do not drive the conversation. I also saw that real-world constraints matter there. Legal, marketing, and consistency of user experience come up a lot, so the best answers are not just clever, they are shippable. And right now there is definitely an expectation that I can think about AI and automation as part of product ideas, not as an afterthought.