Choosing the right metrics is one of the most important responsibilities of a product manager. Metrics shape decisions, influence prioritization, and provide clarity about whether your product is moving in the right direction. But not all metrics are good metrics—and not all numbers tell a meaningful story.
Product managers often fall into common traps: tracking too many metrics, selecting vanity metrics that look impressive but don’t drive action, focusing on outputs instead of outcomes, or measuring things that can’t realistically be influenced. To avoid these pitfalls, PMs must follow a disciplined approach to metric selection.
This article dives into how to pick good metrics. We’ll explore the characteristics of meaningful metrics, explain the difference between exploratory and reporting metrics, and provide practical guidance to ensure you always measure what truly matters.
This is the fourth article in our series, building on earlier discussions about what metrics are, how major companies use them, and the five universal metric categories.
Let’s begin by understanding why choosing the right metrics is such a critical step.
1. Why Choosing Good Metrics Is Harder Than It Looks
Metrics seem simple: pick numbers, track them, analyze them.
But in reality, choosing the right metrics is a strategic exercise with major consequences.
Poor metrics can:
- mislead teams
- drive the wrong decisions
- hide emerging problems
- create false optimism
- waste engineering and UX effort
Good metrics, on the other hand:
- reveal opportunities
- uncover problems early
- provide clarity about performance
- help align teams and stakeholders
- support experimentation and learning
A PM who chooses good metrics builds products intelligently.
A PM who chooses bad metrics flies blind.
2. Two Types of Metrics: Exploratory and Reporting
Before we choose metrics, we need to understand the two broad categories.
a) Exploratory Metrics
Exploratory metrics help you discover what’s happening.
You don’t track them all the time, but you use them when you’re investigating phenomena or forming hypotheses.
Examples:
- click-through rates on an onboarding step
- percentage of users who scroll beyond a certain point
- response time of an API endpoint under load
- number of sessions that result in feature exploration
- heatmap interactions
- video watch completion curves
Exploratory metrics are temporary. They help you answer questions like:
- “Why is activation dropping?”
- “Where are users abandoning their shopping cart?”
- “Which parts of the landing page do users ignore?”
These metrics are used for diagnosis, experimentation, and learning.
They support research more than long-term operations.
b) Reporting Metrics
Reporting metrics are the metrics you track over time.
They provide a stable picture of product performance and are the backbone of dashboards.
Examples:
- monthly active users (MAU)
- retention rate
- churn rate
- NPS
- conversion rate
- MRR / ARR
Reporting metrics are stable, consistent, and comparable period over period.
They must be:
- clearly defined
- easy to interpret
- tied to business or user value
- aligned with product goals
These are the metrics you show leadership.
These are the metrics tied to OKRs.
These are the metrics that define success.
3. The 4 Characteristics of Good Metrics
Great product metrics share four key characteristics.
They must be:
- Understandable
- A Rate or Ratio
- Correlated
- Changeable
Let’s explore each one in detail.
Characteristic #1: Understandable
A good metric should be simple, intuitive, and easy for everyone to grasp: PMs, engineers, designers, marketers, sales teams, and executives.
A metric that requires a three-paragraph explanation is not a good metric.
If stakeholders can’t understand your metric quickly:
- they won’t trust it
- they won’t use it
- they won’t adopt decisions based on it
Examples of understandable metrics:
- “Percentage of users who complete onboarding”
- “Average order value per customer per day”
- “Activation rate: users who sign up and complete their first action”
Examples of confusing metrics:
- “Weighted path progression index”
- “Adjusted rolling engagement heat score”
Complexity hides insight.
A strong PM simplifies.
Characteristic #2: A Rate or Ratio
Metrics that are expressed as raw counts are often misleading.
Rates and ratios create clarity and consistency.
Consider the difference:
- Raw count: “1,000 users signed up this week.”
- Rate: “5% of visitors signed up this week.”
The raw count tells you volume.
The rate tells you effectiveness.
Rates and ratios adjust for:
- changes in traffic
- seasonal variations
- experiments
- user segments
Examples of strong metrics:
- conversion rate
- retention rate
- activation rate
- churn rate
- messages per active user
Ratios give you a fair comparison across time, cohorts, and experiments.
Characteristic #3: Correlated
A good metric must be correlated with what you actually care about.
For example:
You might want users to buy more frequently.
A correlated metric is:
- “number of checkouts per active user”
But measuring:
- “pageviews per user”
is not correlated with purchasing behavior.
Correlated metrics should:
- reflect progress toward a goal
- indicate a real behavioral shift
- change when the underlying behavior changes
Correlation prevents teams from optimizing for meaningless numbers.
Characteristic #4: Changeable
Perhaps the most overlooked characteristic:
A metric must be something your team can influence.
If you can’t change it, you can’t manage it.
Examples of unchangeable (or hard-to-change) metrics:
- “percentage of users who use iPhones vs Androids”
- “number of competitors in the market”
- “GDP growth rate”
- “global search traffic on a specific keyword”
Teams waste time when they try to influence metrics outside their control.
Examples of changeable metrics:
- activation rate
- onboarding completion rate
- cart abandonment rate
- engagement per feature
- NPS (through experience improvements)
Changeable metrics allow teams to take meaningful action.
4. How to Select Good Metrics in Practice
Now that we know the characteristics, let’s apply them to real decision-making.
Here’s a practical approach:
Step 1: Start with the goal
You must be clear about what you want to achieve.
Examples:
- improve retention
- reduce churn
- increase revenue
- improve onboarding
- drive engagement with a new feature
- reduce support tickets
Without a goal, you will pick arbitrary metrics.
Step 2: Define the key behavior that signals success
Ask yourself:
“What must users do for us to achieve this goal?”
Examples:
Goal: improve onboarding
→ key behavior: users complete step 1, step 2, step 3
Goal: improve retention
→ key behavior: users return within a specific period
Goal: increase revenue
→ key behavior: users upgrade or purchase more features
A metric must measure that behavior.
Step 3: Check whether the metric meets the 4 characteristics
Ask:
- Is it understandable?
- Is it a rate or ratio?
- Is it correlated with the goal?
- Is it changeable by the team?
If not, find a better metric.
Step 4: Distinguish whether it is exploratory or reporting
- Exploratory metrics are temporary, for discovery.
- Reporting metrics are long-term, for monitoring.
The right mix depends on your stage and goals.
Step 5: Limit the number of reporting metrics
Most teams should focus on 3 to 7 reporting metrics.
More than that creates noise.
Great PMs optimize for clarity.
5. Common Pitfalls When Choosing Metrics
Even experienced PMs fall into these traps.
Pitfall 1: Choosing vanity metrics
Vanity metrics look impressive but don’t drive behavior or insight.
Examples:
- total pageviews
- total app downloads
- registered users (not activated users)
These create false confidence.
Pitfall 2: Tracking too many metrics
More numbers ≠ more insight.
More numbers often create paralysis.
Focus > breadth.
Pitfall 3: Not validating whether a metric actually changes
Sometimes teams pick metrics that don’t move—even when the product changes.
A static metric is not useful.
Pitfall 4: Measuring outputs instead of outcomes
Outputs measure activity.
Outcomes measure impact.
Example:
- Number of features shipped (output)
- Increase in retention (outcome)
PMs should focus on outcomes.
6. Why Good Metrics Are a Competitive Advantage
Companies that choose good metrics:
- ship better products
- react faster to issues
- align teams effortlessly
- improve user satisfaction
- increase revenue
- create a culture of learning
Choosing the right metrics is one of the highest-leverage skills a PM can develop.
7. Conclusion: The Art and Discipline of Choosing Good Metrics
Picking good metrics requires clarity, intentionality, and discipline.
The best metrics are:
- simple
- rates or ratios
- correlated with real goals
- changeable by your team
They help you monitor performance, diagnose issues, and make strategic decisions.
In the next article of this series, we will explore the HEART Framework, a widely used method for creating structured, customer-centered reporting metrics across happiness, engagement, adoption, retention, and task success.