Balkrishna Attal

Home / Blog / Analytics

AnalyticsSeptember 28, 2026·6 min read

Product Metrics Every PM Should Know: Formulas, Worked Examples and Pitfalls

About twenty metrics cover nearly every product conversation. For each: the formula, a worked number you can check, and how it misleads.

— views
42% 5% 3.3x 30% +30 6 mo
Quick answer

Six families of metrics cover most product work: acquisition, activation, engagement, retention, revenue and quality. Know the formula and the worked example for each — for instance churn = users lost ÷ users at start, and LTV ≈ ARPU × (1 ÷ churn). Choose metrics from your goal, and always pair a number with its definition.

There are hundreds of product metrics, and you will never use most of them. But a shortlist of about twenty shows up in nearly every product conversation, interview and dashboard. This page is the reference: for each, what it means, how it is calculated (with a worked number you can check), and the one way it most often misleads.

All the worked numbers below are illustrative, and every calculation is simple enough to check by hand.

The six families of metrics

AcquisitionHow users arriveActivationFirst real valueEngagementHow much they use itRetentionDo they stay?RevenueDo they pay?QualityHow they feel
A healthy dashboard has one or two metrics from each family, not thirty from one.

Acquisition

MetricFormulaWorked exampleMisleads when
Conversion rateCompleted the action ÷ people who could300 signups ÷ 6,000 visitors = 5%Traffic mix changes; a new low-intent channel drags it down
CAC (customer acquisition cost)Total acquisition spend ÷ new customers₹6,00,000 spend ÷ 500 customers = ₹1,200You forget to include salaries and tools, or count organic users
Traffic by channelVisitors or signups per sourceOrganic 45%, ads 35%, referral 20%Volume is high but quality is low

Activation

Activation is the moment a new user first gets real value, defined by you, for example “completed a first lesson within 7 days”. It is the most important early metric because everything after depends on it.

MetricFormulaWorked exampleMisleads when
Activation rateUsers reaching the key action ÷ new users230 of 500 new users did a first lesson = 46%The “key action” is too easy, so it stops predicting retention
Time to valueMedian time from signup to key actionMedian 2 daysAverages are skewed by a few very slow users; use the median

Engagement

MetricFormulaWorked exampleMisleads when
DAU / WAU / MAUUnique users active in a day / week / monthDAU 12,000, MAU 40,000“Active” is defined loosely (just opening the app)
StickinessDAU ÷ MAU12,000 ÷ 40,000 = 30%Different products have different natural frequencies
Feature adoptionUsers of the feature ÷ eligible users1,800 ÷ 6,000 = 30%The denominator includes people who could never use it

A word of warning on “time spent”: it can rise because users love the product or because they are confused. Use it only alongside a measure of success, as I described in this essay.

Retention and churn

MetricFormulaWorked exampleMisleads when
Day-N retentionUsers active on day N ÷ users who started on day 040 of 100 active on day 7 = 40%You compare it across products with different natural usage
Churn rateUsers lost in the period ÷ users at the start500 ÷ 10,000 = 5% per monthYou mix up user churn and revenue churn
Cohort retention curveRetention by signup group over timeCurve flattening at ~25%You read only the average, not each cohort

The shape matters more than any single number. If the retention curve flattens, you have found a group of users who stay. If it keeps falling towards zero, no amount of acquisition will fix your product. See the cohort table in the product analytics guide.

Revenue and unit economics

MetricFormulaWorked exampleMisleads when
ARPURevenue ÷ users in the period₹200 per user per monthA few large customers hide that most pay little
LTV (simple)ARPU × expected lifetime₹200 × 20 months (1 ÷ 5% churn) = ₹4,000Churn is unstable, so lifetime is a guess
LTV : CACLTV ÷ CAC₹4,000 ÷ ₹1,200 = 3.3xLTV is projected but CAC is actual
CAC paybackCAC ÷ monthly revenue per customer₹1,200 ÷ ₹200 = 6 monthsYou ignore the cost of serving the customer

Worked through: if monthly churn is 5%, a customer stays about 20 months on average, so at ₹200 a month their lifetime value is ₹4,000. If it cost ₹1,200 to acquire them, you earn 3.3 times your acquisition cost and take about 6 months to recoup it. A common rule of thumb is that a ratio above about 3 is healthy, but treat it as a prompt for questions, not a verdict, because both numbers are estimates.

Quality and satisfaction

MetricFormulaWorked exampleMisleads when
NPS% promoters (9-10) minus % detractors (0-6)50% promoters − 20% detractors = +30Response rate is low or biased toward happy users
CSATSatisfied responses ÷ total responses180 of 240 rated 4 or 5 = 75%It measures one moment, not the whole relationship
Error / crash rateSessions with an error ÷ all sessions0.8% of sessionsRare but severe errors are hidden in the average

Experiment metrics

ConceptMeaningWatch out for
Primary metricThe one number the test is designed to moveChoosing it after seeing results
Guardrail metricA number that must not get worseIgnoring it because the primary won
Lift(Variant − control) ÷ controlReporting lift without saying how big the sample was
Statistical significanceEvidence the difference is not just noiseStopping early when it briefly looks significant

Which metrics should you actually track?

Activation rateearly healthDay-7 / day-30 retentiondo they stayConversion to paiddo they payStickiness (DAU/MAU)habit strengthNPS or CSAThow they feel
An illustrative priority order for a young consumer product. Yours may differ; start from your goal, not this list.

Start from the goal and work backwards. Early on, activation and retention tell you whether the product works at all. Once retention is healthy, look at conversion and unit economics. Add satisfaction measures alongside, not instead. If you track a metric and no decision would change whether it goes up or down, stop tracking it.

Common mistakes with metrics

Reporting a metric without its definition. Using an average where a median or a distribution is needed. Watching totals (which only go up) instead of rates. Comparing periods of different length or cohorts of different age. Optimising a single number until it breaks another. And treating a benchmark from another company as your target, when your users, price and category are different.

Keep going: product analytics, SQL for PMs, choosing a North Star, and the metrics section of the 50 question bank.

Frequently asked questions

What product metrics should every PM know?

Conversion rate, CAC, activation rate, DAU/WAU/MAU and stickiness, day-N retention, churn, ARPU, LTV and LTV:CAC, NPS or CSAT, and experiment concepts like primary and guardrail metrics.

What is a good LTV to CAC ratio?

A commonly cited rule of thumb is around 3 or higher, but it depends on the business, margins and how reliable your LTV estimate is. Treat it as a prompt for questions, not a verdict.

How do you calculate churn rate?

Divide the number of users lost during a period by the number of users at the start of that period. For example, losing 500 of 10,000 users in a month is a 5% monthly churn rate.

What is the difference between retention and churn?

Retention is the share of users who stay; churn is the share who leave. Over the same period they add up to 100%, so 95% monthly retention is 5% monthly churn.

Want this sharpened for your specific situation?

I take on a small number of career calls and resume reviews each month.

See how we could work together
BA

Balkrishna Attal

A product leader with a background in civil engineering. Writes about product management careers, interviews, and the frameworks that hold up in real work.