A pragmatic product leader is constantly measuring. Always has major numbers in mind and the numbers that he cares about.
Metrics are not an afterthought. They are the oxygen that fuels your product decisions and the compass that guides your team’s efforts. The trap most PMs fall into is tracking too many numbers — metrics for the sake of metrics — without understanding which ones actually move the needle.
The actual job is to pick metrics that reveal whether you are progressing toward your goals, and to use them to steer product development, not just report on past activity.
Metrics create momentum — the formula is simple
Think of momentum in physics: momentum equals mass times velocity (M = mv). In product terms, mass is the metric — the number that matters — and velocity is the speed at which it moves toward your goal.
For example, say your goal is to improve retention by 10% over a month. Your metric is retention. Your velocity is the weekly rate of improvement needed to hit that target — about 2.5% per week.
This is how you quantify progress. Without metrics, your theories and efforts are just guesses.
The difference between KPIs and metrics
Metrics are all the numbers you track — daily active users, clicks, session length. KPIs (Key Performance Indicators) are a subset of metrics that directly measure success against your strategic objectives.
KPIs are your navigational instruments. They are:
- Actionable: They guide decisions and next steps.
- Aligned: They reflect company and product goals.
- Measurable: You can track them reliably over time.
For example, Swiggy tracks delivery time as a KPI because it directly affects customer satisfaction and repeat orders. Measuring app installs alone is not enough if users uninstall quickly.
The trap of too many metrics
It is easy to get overwhelmed by metrics. Every feature, every user action generates data. But most of it is noise.
Talvinder says: "Picking good metrics means spotting the signal from the noise." You want to measure only what helps you improve the product next sprint or release.
For example, if you want to improve search experience, you might track dozens of signals — query correctness, suggestion relevance, response time, click-through rates, etc. But start by listing all hypotheses about what could be wrong, then prioritize metrics that test the biggest suspects.
If your search conversion rate is half the industry average, you might hypothesize that irrelevant results are the culprit. You test this by tracking click-through on top results. If that’s low, you target improving relevance.
Product analytics review meeting at a Series A Indian e-commerce startup
Product Lead: “Search conversion is at 10%, industry average is 20%. We need to fix this.”
Data Scientist: “We found that click-through on top 3 results is just 15%.”
Product Lead: “That confirms relevance is a problem. Let’s prioritize improving search ranking algorithms.”
Engineering Lead: “We’ll start A/B testing new ranking models next sprint.”
Choosing the right metric to focus on can make or break your next release.
Frameworks to pick the right metrics
You cannot pick good metrics without a framework. Two frameworks Talvinder recommends are HEART and AARRR.
HEART framework (Google Ventures)
| Letter | Meaning | What it measures | Example metrics |
|---|---|---|---|
| H | Happiness | User satisfaction and sentiment | NPS, CSAT, user surveys |
| E | Engagement | Frequency and depth of use | Visits per session, search per session |
| A | Adoption | New users starting to use the product or feature | % of new signups using a feature |
| R | Retention | Users returning after first use | % retained after 1 week, 1 month |
| T | Task success | How well users complete core tasks | Task completion rate, error rate, time on task |
HEART helps you cover the user experience comprehensively. Use it to brainstorm metrics that matter for a feature or product.
AARRR framework (Pirate metrics)
| Letter | Meaning | What it measures | Example metrics |
|---|---|---|---|
| A | Acquisition | How users find your product | Website visitors, app installs |
| A | Activation | Initial user experience success | Signup to first purchase conversion |
| R | Retention | Repeat usage over time | Monthly active users |
| R | Revenue | Monetization success | Average revenue per user |
| R | Referral | How users spread the word | % users who invite friends |
AARRR focuses on the customer lifecycle and is especially useful for growth and funnel analysis.
Input metrics vs output metrics
Metrics fall into two broad categories:
-
Input metrics: These are leading indicators — actions or behaviors that drive outcomes. For example, number of product searches, number of messages sent, number of accounts created.
-
Output metrics: These are lagging indicators — results or outcomes. For example, revenue, retention rate, customer satisfaction.
A pragmatic PM tracks input metrics to diagnose problems early and output metrics to measure ultimate success.
The trap is to optimize output metrics without understanding what drives them. For example, increasing revenue by increasing prices might hurt retention. Tracking both input and output metrics together is essential.
Traceability: linking metrics to product decisions
Metrics should not be isolated numbers. They must be traceable — meaning you can connect a metric back to specific product features, experiments, or changes.
If you launch a new search algorithm, you should track how it impacts search success rate, conversion from search, and ultimately retention.
If a metric changes, you should be able to explain why — which experiments or changes caused it.
Talvinder explains:
"As a product manager, your job is to measure the impact of your work. If you cannot tie a metric change to a product change, you’re not really managing the product — you are just watching numbers."
The Indian context: what metrics matter here?
India’s market nuances influence which metrics matter:
-
Mobile-first users: Metrics like app crash rate, load time on 2G/3G networks, and battery usage are critical.
-
Diverse languages: Engagement metrics segmented by language and region reveal gaps in localization.
-
Payment preferences: For fintech, UPI transactions and wallet top-ups are key KPIs.
-
Price sensitivity: Conversion and retention metrics vary greatly with pricing tiers and discount campaigns.
For example, Flipkart found improving app performance in tier-2 and tier-3 cities led to a significant lift in conversions. Meesho’s referral and activation metrics differ by region because of social selling dynamics.
Common pitfalls in metric selection
-
Vanity metrics: Metrics that look good but don’t correlate to business goals. Example: total app downloads without retention.
-
Tracking everything: Leads to analysis paralysis and no clear focus.
-
Ignoring qualitative signals: Metrics don’t tell you why users behave a certain way.
-
No baseline or target: Without a benchmark, metrics have no meaning.
-
Not updating metrics: As product and business evolve, so should your KPIs.
How to decide what to cut down
When overwhelmed, ask:
- Does this metric help us decide what to build next?
- Can we measure it reliably and frequently?
- Does it correlate with user value or business success?
- Is it actionable by the product team?
If the answer is no, drop it.
FieldExercise title="Select your product’s north star metrics" time="15 min"
Pick a product you work on or use regularly.
- Write down its core user problem and business goal.
- Using HEART or AARRR, list 3-5 candidate metrics.
- Mark which are input vs output metrics.
- For each metric, write how it connects to a product decision or experiment.
- Identify 1-2 vanity metrics and decide whether to keep or drop them.
- Define a target or benchmark for each key metric.
JudgmentExercise
scenario="You are a PM at a Series B Indian fintech startup. Your team launched a new feature to improve loan application approval rate. After rollout, your dashboard shows loan applications increased by 20%, but approval rate dropped by 10%, and customer complaints doubled. You have multiple metrics to analyze." question="Which metrics do you prioritize to investigate the problem and what actions do you recommend?" expertReasoning="Prioritize approval rate and customer complaints as lagging output metrics indicating quality issues. Investigate input metrics like application completeness, fraud detection alerts, and processing time to diagnose causes. Recommend halting the feature rollout if approval falls below acceptable thresholds. Use traceability to link metric changes to feature components and iterate with targeted fixes." commonMistake="Focusing solely on the increase in loan applications (a vanity metric) without considering approval quality and customer satisfaction. This leads to pushing a flawed feature that harms the product’s reputation and long-term retention." />
You are a PM at a Series B Indian fintech startup. Your team launched a new feature to improve loan application approval rate. After rollout, your dashboard shows loan applications increased by 20%, but approval rate dropped by 10%, and customer complaints doubled. You have multiple metrics to analyze.
Your task: Which metrics do you prioritize to investigate the problem and what actions do you recommend?
your reasoning:
FromTheField context="from a Pragmatic Leaders AMA on metrics"
I have watched thousands of PMs struggle with metrics. The common thread is this: they track too many numbers without a clear hypothesis. The result is confusion and poor prioritization. Good PMs start with the goal, pick one or two metrics that reflect that goal, and relentlessly focus on improving them. When you measure what matters, you can move fast and with confidence.
Where to go next
- Master user research to complement metrics: User Research Methods
- Learn how to translate metrics into product vision: Product Vision and Strategy
- Understand how to measure feature success: Metrics and KPIs
- Prepare for product interviews with metrics questions: PM Interviews
PL alumni now work at Razorpay, Meesho, Swiggy, Flipkart, PhonePe, and dozens of other Indian startups.