Chapter 4 · Product Analyst
4. Funnels, cohorts & retention
~9 min read
These three analyses are the bread and butter of product work. Together they answer: where do users drop off, how do different groups behave over time, and does the product actually keep people?
4.1 Funnel analysis#
A funnel breaks a journey into ordered steps and measures conversion between them. Its job is to find the biggest leak, the step where you lose the most users relative to its importance.
| Step | Users | Step conversion |
|---|---|---|
| Visit | 10,000 | , |
| Sign up | 4,200 | 42% |
| Activate | 2,800 | 67% |
| Purchase | 980 | 35% |
| Repeat | 520 | 53% |
Best practices#
- Define the window. Does a user have one session or seven days to convert?
- Strict vs. loose ordering. Must steps happen in sequence, or just all occur?
- Watch time-to-convert. A step users eventually pass but take days on is friction.
- Segment. The leak is often specific to one platform, channel, or user type.
4.2 Cohort analysis#
A cohort is a group of users bound by a shared start, usually signup week. Cohort analysis tracks each group over time so improvements aren't hidden by the constant inflow of new users.
4.3 Retention, the truest measure of PMF#
A healthy product's retention curve flattens to a stable plateau, a core of users who keep getting value. A failing product's curve decays toward zero.
| Days since signup | Healthy (flattens) | Leaky (decays) |
|---|---|---|
| 0 | 100 | 100 |
| 1 | 62 | 55 |
| 7 | 40 | 22 |
| 14 | 37 | 14 |
| 30 | 33 | 8 |
| 60 | 31 | 5 |
| 90 | 30 | 3 |
| Retention type | Counts a user as retained if they… | Best for |
|---|---|---|
| N-day | Return on exactly day N | Products with a daily habit |
| Unbounded / rolling | Return on day N or any day after | Most products; less noisy |
| Bracket | Return within a window (e.g. days 5, 7) | Weekly-use products |
4.4 Reading a cohort retention table#
| Cohort | Wk0 | Wk1 | Wk2 | Wk4 |
|---|---|---|---|---|
| Jan W1 | 100% | 41% | 33% | 28% |
| Jan W2 | 100% | 44% | 36% | 31% |
| Jan W3 | 100% | 48% | 40% | , |
| Jan W4 | 100% | 52% | , | , |
Get the next chapter and weekly interview tips by email
One short email per week. Skim in a minute. Unsubscribe anytime.
