Product Analytics
for Product Managers (2026)
Retention curves, funnels, A/B tests, cohort analysis, and event tracking — the analytics concepts every PM needs, explained without the data science jargon.
Practice Analytics Questions Daily — Free →Retention Curves
How many users from a given cohort are still active N days later. The most honest signal of product health.
How to read it
Day 1 retention = % who return the day after signup. Day 7 = week 1 habits. Day 30 = product stickiness. A healthy product curve flattens — not drops to zero.
PM use case
Compare retention by acquisition channel, user segment, or feature usage. Users who use Feature X have 2x D30 retention → Feature X is a habit driver.
⚠️ Red flag to watch: If D7 retention is under 20% for a consumer app, the core loop is broken. Fix that before anything else.
SELECT cohort_date, COUNT(DISTINCT user_id) as retained FROM events WHERE event_date = cohort_date + INTERVAL 7 DAY GROUP BY cohort_date
Funnel Analysis
The conversion rate through each step of a user journey — from landing to activation, or from cart to purchase.
How to read it
Each step has a drop-off rate. The step with the biggest drop-off is the highest-leverage improvement opportunity.
PM use case
Run weekly funnel reviews. When a step drops unusually, investigate immediately — it's usually a bug, a UX change, or an external factor.
⚠️ Red flag to watch: A step that shouldn't require effort (e.g. 'Continue to next step') with >30% drop-off usually means there's a bug or severe UX friction.
Count distinct users at each event in sequence. Divide each step by the step before it to get conversion rates.
A/B Testing
Running two versions of a feature simultaneously on randomly split user groups to measure which performs better.
How to read it
Statistical significance (p < 0.05) means the result is unlikely to be random chance. Effect size tells you if it's worth shipping.
PM use case
Always pre-register your hypothesis before seeing results. Post-hoc hypothesis generation leads to false positives. Run tests long enough to capture weekly cycles.
⚠️ Red flag to watch: Peeking at results too early and stopping when you see significance is p-hacking. Wait for the pre-determined sample size or time period.
Use a chi-squared test or t-test for proportions. Most tools (Amplitude, Optimizely) calculate this for you — but know what the number means.
North Star Metric
The single metric that best captures the value your product delivers to users — and is a leading indicator of long-term business health.
How to read it
A good north star moves when users get value and leads revenue growth. A bad north star can be gamed, lags too far behind user behaviour, or conflates different user journeys.
PM use case
Decompose your north star into input metrics (things teams can directly influence). Build your roadmap around moving input metrics.
⚠️ Red flag to watch: If your north star is revenue or an output metric — you're measuring outcomes, not causes. You'll ship things that move revenue short-term while destroying long-term health.
Define the event that counts as 'value delivered' and count unique users or sessions where that event occurred per day/week.
Cohort Analysis
Grouping users by when they joined (or did a specific action) and tracking their behaviour over time separately per group.
How to read it
If January cohorts retain better than October cohorts, something changed between Jan and Oct — product change, acquisition channel shift, or seasonality.
PM use case
Use cohort analysis to measure the real impact of product changes over time. Aggregate metrics average over all cohorts and can hide trends.
⚠️ Red flag to watch: Looking only at aggregate DAU and missing that newer cohorts retain much worse. This is how teams miss the early signal of product degradation.
Add a signup_date to every user record. Group events by FLOOR(DATEDIFF(event_date, signup_date) / 7) for weekly cohort retention.
Event Tracking Schema
The structured naming and data model for the events your product emits — what gets tracked, what properties are attached, and how they're named.
How to read it
Good schema: verb_noun format (button_clicked, lesson_completed). Bad schema: inconsistent names, missing user_id, or no timestamp.
PM use case
PMs should own the event tracking plan. Define events before engineering builds the feature — not as an afterthought. Missing events can never be retroactively captured.
⚠️ Red flag to watch: Discovering post-launch that you can't answer 'how many users completed step 3?' because you never tracked step 3. This is a planning failure, not an analytics failure.
Every event should have: user_id, session_id, event_name, timestamp, and relevant properties. Define these in a tracking spec before sprint starts.
PM Analytics Tools to Know
FAQ
How much analytics do product managers actually need to know?
Enough to be self-sufficient on routine questions (what's my retention, where's the funnel drop, did this A/B test win?) and enough to have an intelligent conversation with your data analyst on deeper analysis. You don't need to build dashboards from scratch or write complex SQL — but PMs who can write basic SQL and read a p-value make better, faster decisions.
What analytics tools should a PM know in India?
Amplitude and Mixpanel for product analytics are the standard at most Indian startups and growth-stage companies. Basic SQL (Postgres or BigQuery syntax) is increasingly expected at data-driven companies. Google Analytics 4 for web products. The most important skill isn't the tool — it's knowing what question to ask and whether the data you're looking at actually answers it.
Build Analytics Intuition in 2 Minutes a Day
Daily PM analytics scenarios — interpret real dashboards, diagnose metric drops, design experiments.
Start Free Trial →