Launching a product creates a specific measurement problem that general metrics frameworks don't fully address. When a feature is established, you have baselines, historical patterns, and comparison cohorts. At launch, you have none of those. You're measuring something that has never existed before, in a market that hasn't responded yet, with users who don't yet know what they think.

The candidates who answer "how would you define success for this launch?" well understand that distinction. They know which metrics are predictive of long-term success in the first 30 days, which metrics are too slow to be useful at launch, and what instrumentation you need before you go live rather than after.

Why Launch Metrics Are Different

For a live product, your goal is to detect change against a known state. A metric drop of 10% week-over-week is alarming because you know what normal looks like. See the PM Metric Drop Root Cause Analysis guide for how that diagnostic process works.

At launch, you don't have a "normal" yet. You're trying to establish whether early signals indicate the product has traction, not whether something broke. The questions you're asking are different:

Are the right users finding this?
Are they doing what we designed them to do?
Are they coming back?

You're also asking these questions before you have enough data to answer them with statistical confidence. This is why launch measurement requires a different discipline: tracking leading indicators that correlate with long-term success, not just the lagging indicators that confirm it.

Leading vs. Lagging Indicators at Launch

Lagging indicators are the outcomes you ultimately care about: revenue, monthly active users (MAU), net promoter score (NPS). They're the right metrics to optimize for over time, but they're slow to move and slow to interpret. A revenue figure three months post-launch reflects decisions made in weeks one and two. By the time it confirms you had a problem, it's too late to course-correct early.

Leading indicators are the behaviors that predict whether lagging outcomes will follow. They move faster and give you signal while there's still time to act.

Examples of useful leading indicators at launch:

Activation rate. The percentage of new users who complete the critical first action that correlates with retention. If activation in week one is low, retention in week four will be low. You know this before week four arrives.
Time-to-first-value. How long does it take a new user to experience the core value of the product? If this is 45 minutes, most users will churn before they get there.
Day-1 and day-7 retention. These don't take 90 days to compute. Day-1 retention tells you whether users found the initial experience worth returning for. Day-7 tells you whether there's a habit forming.
Feature adoption depth. What percentage of activated users have used more than one feature? Depth predicts stickiness better than breadth of acquisition.
Session quality signals. Error rates, task completion rates, and time-on-task tell you whether the product is working technically and whether users can accomplish what they came to do.

The general success metrics framework for measuring an established product applies here too. The launch version simply prioritizes earlier-signal metrics. See How to Measure Product Success in PM Interviews for the full framework.

Instrumentation You Need Before Go-Live

One of the most common PM mistakes at launch is discovering you don't have the data you need after the launch has already happened. Instrumentation planning is a pre-launch task, not a post-launch one.

Before launch, confirm you can measure:

The activation event. What is the specific user action that constitutes "activation" for this product? Is it tracked? Does it fire correctly in your event logging system?
The core loop. Can you track whether users who activated return to repeat the core action? Retention calculation requires both an initial event and a return event.
Funnel steps. If users go through a multi-step flow, is every step instrumented? Funnel drop-off is invisible if intermediate steps aren't logged.
Error states. Are errors surfaced in your logging system? Are they tagged to specific features or flows? You need to know when users hit walls.
Attribution. Where are users coming from? If you're running acquisition campaigns alongside the launch, you need source tagging to know which channels are producing retained users vs. one-time visits.

A launch without instrumentation is not just a measurement failure. It prevents learning. If something goes wrong and you have no event data, you can't diagnose it. If something goes right and you don't know why, you can't replicate it.

The pre-launch instrumentation checklist is worth building as a shared artifact with your engineering team. Not every company has a formal QA (quality assurance) process for analytics, but establishing one for launches is high-leverage.

Setting a Baseline When You Have No History

At launch, you have no historical baseline for your own product. You have three substitutes:

1. Internal analogues. If your company has launched similar products or features before, use those as reference points. "Our last major feature launch reached 20% activation in the first two weeks" is a weak baseline but better than nothing. Document what was similar and what was different about the context.

2. Industry benchmarks. SaaS (software as a service) products typically see day-30 retention rates in the 20-40% range for consumer apps and 60-80% for enterprise tools. Mobile app benchmarks vary by category. These numbers are noisy and context-dependent, but they tell you whether your early numbers are in the plausible range.

3. Pre-launch research signals. Beta cohort data, waitlist conversion rates, and early access user behavior can provide predictive signals. If 40% of beta users activated and 60% returned in week two, you have a hypothesis about what launch performance might look like, and a benchmark to compare against.

Be explicit about which substitute you're using and what its limitations are. In an interview, saying "I'd use industry benchmarks for day-7 retention as a reference point, while acknowledging they're noisy, and set a watch threshold based on whether we're in the bottom quartile rather than trying to hit a precise number" is a stronger answer than claiming you'd have a perfect baseline.

What a Good Launch Dashboard Looks Like

A launch dashboard is not a general analytics view. It's a focused instrument for the first 30 to 90 days, designed to answer a specific question: is this product developing the right signals to predict long-term success?

What to include:

A launch dashboard should have four sections.

Acquisition. New users by source and channel. Not just total users: you need to know which sources are producing users who activate and retain, not just sign up. A paid campaign that drives 10,000 signups with 5% activation is worse than an organic channel that drives 500 signups with 60% activation.

Activation. The activation rate, broken down by channel and cohort. If activation rates differ significantly between acquisition sources, that tells you something about who the product is working for.

Retention. Day-1, day-7, and day-30 retention curves. Updated daily for new cohorts. The earlier you can see retention curves flattening (users stabilizing around a return frequency), the earlier you can confirm product-market fit is developing.

Health. Error rates, latency, crash rates, and customer support volume. These are guard rails. If a retention curve looks good but support tickets are climbing, something is working for some users and breaking for others.

What to exclude:

Vanity metrics. Total downloads, total page views, and total registrations tell you about reach, not traction. A launch dashboard filled with large numbers that don't correlate with retention creates false confidence.

Failed Launch vs. Slow Start: How to Tell the Difference

This is the hardest interpretive judgment at launch. The signals for "this product isn't working" and "this product needs more time to find its market" can look similar in week three.

Signals of a failed launch:

Activation rate is below the bottom quartile of analogues and isn't improving as you iterate on onboarding.
Users who do activate churn immediately. Day-1 retention is under 20% and not improving.
Qualitative feedback is confused, not disappointed. Users who churn without completing activation didn't understand what the product does, not just whether they want it.
No segment is responding. In a genuine slow start, you typically see one cohort or segment showing strong signals even if overall numbers are weak. A flat signal across all segments suggests the core value proposition isn't landing for anyone.

Signals of a slow start:

A specific segment is showing strong leading indicators. The product is working for someone. The question is whether that segment is large enough.
Activation rate is low but improving week-over-week as you refine the onboarding.
Users who do activate and return are deeply engaged. If your core retention cohort is small but has high session frequency and feature depth, you have a product that works for a narrow audience. That's a positioning and acquisition problem, not a product problem.
Qualitative feedback is specific. Users who churn can articulate what they wanted that they didn't get. That's a solvable problem.

The practical threshold: if, after 8 to 12 weeks, no segment shows strong activation and day-7 retention, and qualitative research isn't revealing a solvable gap, that's a failed launch signal. If the product is iterating toward a working onboarding and retaining any engaged cohort, you're looking at a slow start that warrants continued investment.

How to Answer This in a PM Interview

When an interviewer asks "how would you define success for this product launch?", they're testing three things:

Whether you distinguish leading from lagging. Candidates who answer with "monthly active users and revenue" are giving the right long-term metrics but missing the launch-specific framing. Acknowledge that MAU and revenue are the ultimate outcomes, then pivot to what you'd track in weeks one through four to predict whether those outcomes are developing.

Whether you think about instrumentation. Strong candidates bring up what needs to be tracked before launch, not just what they'd look at after. "Before we go live, I'd confirm we have activation events firing correctly, funnel instrumentation for the onboarding flow, and source tagging on all acquisition channels" signals production experience.

Whether you can distinguish failure from early signal. Give your interpretation framework. How would you tell, in week six, whether to double down or cut losses? Having a clear decision rule, even an imperfect one, shows that you've thought about this as a real operational decision and not just a measurement exercise.

For how A/B testing fits into the post-launch iteration cycle, see A/B Testing for Product Managers. For how launch metrics connect to PRD success criteria, see How to Write a PRD.

Cala Voice lets you practice launch metrics questions and success definition questions out loud, with an AI interviewer that pushes back on vague metric choices and asks you to distinguish leading from lagging indicators in real time.

Start your preparation →

How to Define Success Metrics for a Product Launch