How to Read Retention Rate
Retention is one of the most important signals for whether a product has value.
But many founders read retention as a single number:
Day 7 retention: 18%
Day 30 retention: 6%
Then they jump to a conclusion:
Retention is bad
The product is not working
Growth is meaningless
That is not enough.
Retention is not an isolated number.
You need to understand:
Why users return
When users return
Which users return
Which actions make users return
Whether the retention curve stabilizes
What Retention Means
Retention measures how many users from a group continue to come back after a period of time.
Common versions:
Day 1 retention
Day 7 retention
Day 14 retention
Day 30 retention
Example:
100 new users arrived on January 1
20 came back on January 8
Day 7 retention = 20%
But the key question is:
What did they come back to do?
If they only opened the page, that may not be meaningful retention.
Better retention should connect to core actions:
Generate another report
Create another task
Search another keyword
Export another file
Invite another teammate
Complete another learning session
Use Cohorts
A cohort is a group of users who started in the same period.
Do not mix all users together.
Group users by start time:
Users who joined in week 1
Users who joined in week 2
Users who joined in week 3
Then compare whether each group comes back.
This shows whether the product is improving.
Example:
Week 1 users D7 retention: 10%
Week 2 users D7 retention: 14%
Week 3 users D7 retention: 21%
Even if total user count is still small, product quality may be improving.
On the other hand, if new users increase but each cohort retains worse, growth may only be pouring more people into a leaky funnel.
Different Products Need Different Windows
Do not force every product to focus on D1, D7, and D30.
Usage frequency should decide the retention window.
Use this as a guide:
High-frequency products: D1, D3, D7
Medium-frequency products: D7, D14, D30
Low-frequency products: monthly retention, reuse cycle, post-task return
Event-based products: reuse when the next event happens
For a daily vocabulary product, D1 retention matters.
For a contract generation tool, users may not return the next day.
It should track:
Did the user finish the contract?
Did the user save a template?
Did the user generate another contract?
Did the user recommend it to a colleague?
Did the user return during the next contract scenario?
Read the Curve, Not One Point
The most important thing is not one day’s number.
It is the shape of the retention curve.
Common patterns:
Continuous decline: users keep leaving and no value is forming
Sharp drop then stable: unfit users leave, real users remain
Improving cohorts: product, positioning, and activation are getting better
For an early product, the goal is not always extremely high Day 1 retention.
It is more important to see whether the curve stabilizes.
If a small group still uses the product after 30 days, there may be a real value point.
Segment Retention
Overall retention only tells you that there may be a problem.
Segmentation tells you where the problem is.
At minimum, split by:
Channel
New vs returning users
Activated vs unactivated users
Paid users
Use case
Core action
Example:
Search traffic has high D7 retention
Short-video traffic has low D7 retention
That may be a channel quality issue, not a product issue.
Another example:
Users who completed the core action have 35% D7 retention
Users who did not complete it have 5% D7 retention
This means the activation flow is the first thing to improve.
Diagnose Low Retention
Use this sequence:
1. Is the active definition correct?
2. Did users complete the core value action?
3. Do new users understand the product?
4. Is there a reason to return?
5. Are reminders and touchpoints useful?
6. Does the channel match the product?
Do not add features immediately when retention is low.
Often the issue is not a lack of features.
Users simply did not reach the value point.
A Minimal Retention Dashboard
For an early product, start with:
New users
Activated users
D1 retention
D7 retention
D30 retention
D7 retention for users who completed the core action
D7 retention for users who did not complete the core action
D7 retention by channel
Every week, answer:
Which users stayed, and why did they stay?
That is more useful than reporting that retention rose by two percentage points.
Summary
Retention measures whether a cohort of users keeps coming back
Retention should connect to core actions, not page opens
Use cohorts, or you cannot tell whether the product is improving
Different product frequencies need different retention windows
The curve matters more than one number
Segment by channel, activation, paid users, and core actions
When retention is low, inspect value delivery before adding features
Next, we will cover what PMF means.
评论
0分享你对这一章的想法