How to measure feature adoption rate
A thousand banner clicks can hide a feature that hardly anyone used. Count the people who did the thing the feature was built for, then see who did it again.
Pick the first action that delivers value
For a school messaging tool, count a teacher who successfully sends a message. Opening the composer tells you someone found it. A successful send tells you they used it. Count each teacher once during the week, even if they sent messages to three classes.
Who belongs in the denominator?
Feature adoption rate is the share of eligible active users who used the feature during the same period. Suppose 500 teachers were active last week, 300 had access and 60 of those sent a message. Adoption among teachers who could use it was 20%. Across all active teachers, reach was 12%. Use the first number to improve the feature and the second to plan a wider rollout.
| Last week’s question | Calculation | Result |
|---|---|---|
| How many eligible teachers used it? | 60 of 300 | 20% adoption |
| How many active teachers did it reach? | 60 of 500 | 12% reach |
Amplitude’s feature engagement guide shows how to compare the share of active users who use a feature with how often they use it.
Did the first users come back?
Take the 60 senders from that week. If 15 of those same teachers send again the next week, weekly reuse is 25%. A low first-use rate gives you a reason to check access and discovery. If teachers try it once and stop, watch what happens after their first send before promoting it more widely.
What Clever learned from a messaging launch
In prior in-house work at Clever, the team used role and behavior data to target the messaging launch. The case reports that adoption rose 10% among targeted users. The starting rate was not recorded, so the change cannot be restated in percentage points. Weekly retention was 40% among people who engaged with the messaging surface. Read the Clever case study.
To find out whether a reminder caused more people to use the feature, test it with a holdout group.