Metrics for PMs: A Guide to Measuring Success

TL;DR

PM interviews test metric selection, not metric creation. The best candidates frame success around business impact, not vanity numbers. Empty metrics signal weak judgment—hiring committees dismiss them instantly.

Who This Is For

Mid-level PMs interviewing at FAANG who keep getting feedback like “metrics feel arbitrary” or “no clear tie to outcomes.” You’ve shipped features but struggle to articulate how they moved the needle. This is for the candidate whose interviewers say, “I don’t see the judgment here.”

How do you pick the right metric for a PM interview question?

The right metric isn’t the one you can measure—it’s the one that forces a tradeoff. In a Q3 debrief at Google, a candidate proposed “daily active users” for a new search feature. The hiring manager killed the loop: “That’s a lagging indicator. What’s the leading signal that tells us we’re on track or off?” The candidate pivoted to “query refinement rate,” which exposed a tension between relevance and latency. That tension is the point.

Judgment signal: not the metric itself, but the constraint it reveals. Weak candidates pick metrics that only go up. Strong candidates pick metrics that create debate.

Why do interviewers reject vanilla metrics like “user growth”?

Because growth is a given, not a choice. In a Meta debrief, a candidate used “MAU” to justify a notification system. The HC lead responded, “MAU is the company’s metric, not yours. What’s the cost of this feature to retention?” The candidate had no answer. The problem wasn’t the metric—it was the lack of ownership.

Not X: “We’ll track adoption.”

But Y: “We’ll track adoption, but we’ll cap notifications at 3 per day to protect long-term engagement.”

What’s the difference between a North Star and an interview metric?

North Stars are for roadmaps. Interview metrics are for decisions. In a Stripe loop, a candidate proposed “payment success rate” as the North Star for a new checkout flow. The interviewer asked, “What’s the tradeoff between success rate and fraud detection latency?” The candidate’s silence ended the round.

Insight layer: Interview metrics must create a fork in the road. If your metric doesn’t force a prioritization call, it’s empty.

How do you handle metrics for ambiguous products?

Ambiguity is the test. In an Airbnb debrief, a candidate was asked how to measure a new “experience” category. They proposed “bookings per host.” The HM countered, “That incentivizes hosts to spam listings.” The candidate revised to “guest satisfaction score, weighted by repeat bookings.” That revision exposed the real work: balancing supply and demand.

Not X: “We’ll measure what’s easy.”

But Y: “We’ll measure what’s hard, because that’s where the judgment lives.”

When should you abandon a metric during an interview?

When it stops telling a story. In a Twitter loop, a candidate stuck to “tweet volume” for a new composer feature. The interviewer said, “Volume is up, but quality is down. How do you know?” The candidate had no quality proxy. The metric was empty because it didn’t account for its own failure mode.

Judgment call: If your metric can’t explain a downside scenario, it’s not a metric—it’s a blind spot.

Preparation Checklist

List the 3 metrics your last feature should have moved, and the 3 it accidentally broke. Debate the tradeoffs aloud.
For each product question, define a guardrail metric that caps the upside of your primary metric.
Practice articulating the cost of your metric: “If X goes up, Y goes down, and we accept that because Z.”
Work through a structured preparation system (the PM Interview Playbook covers metric tradeoff frameworks with real debrief examples).
Prepare a “metric failure” story: a time your metric lied, and how you corrected it.
For ambiguous products, default to ratio metrics (e.g., engagement per session) over absolute ones.
Never propose a metric you can’t tie to a business outcome in 2 sentences.

Mistakes to Avoid

BAD: “We’ll track user engagement.”

GOOD: “We’ll track engagement per session, but we’ll monitor churn for users who hit the engagement cap.”

BAD: “The metric is revenue.”

GOOD: “The metric is revenue per active merchant, because we need to ensure we’re not just onboarding low-value accounts.”

BAD: “We’ll measure adoption.”

GOOD: “We’ll measure adoption among power users, because casual users will skew the data without driving retention.”

FAQ

How many metrics should you propose in a PM interview?

One primary, one guardrail, one cost. Three is the minimum to show judgment. More than five signals indecision.

What’s the fastest way to kill your metric answer?

Propose a metric that only goes up. Interviewers assume you haven’t thought about the downside.

Can you reuse the same metric across different interview questions?

No. Each product scenario demands a unique metric tension. Reusing metrics proves you’re pattern-matching, not thinking.

Ready to build a real interview prep system?

Get the full PM Interview Prep System →

The book is also available on Amazon Kindle.