Feature adoption: how to measure it and raise it
7 min read
Feature adoption is how many of your users start using a feature after it ships, and how many keep using it. The usual measure is the number of users who used the feature in a period divided by the number who could have, tracked for weeks after launch, because the first few days mostly measure curiosity. When adoption is low, the cause is usually one of three things: people do not know the feature exists, they cannot find it where they work, or it does not solve a problem they have.
| Metric | How to calculate it | What it tells you |
|---|---|---|
| Adoption rate | Users who used the feature ÷ eligible active users, in a period | How far the feature has reached |
| Time to first use | Median days from release (or signup) to first use | How quickly people discover it |
| Feature retention | Users still using it after N weeks ÷ users who tried it | Whether it stuck after the first try |
| Frequency | Uses per active user per week | How central it is to their work |
| Adoption by segment | Adoption rate per plan, role or company size | Who it is actually for |
What is feature adoption?
It is the point at which a user has started using a feature as part of their work, as opposed to seeing it once. Counting it well means deciding what “using” means for that feature, which population could have used it, and over what period.
Google’s HEART framework, from Rodden, Hutchinson and Fu’s CHI 2010 paper Measuring the User Experience on a Large Scale, names Adoption as one of five categories alongside Happiness, Engagement, Retention and Task success. It says adoption metrics “track how many new users start using a product during a given time period,” and that adoption and retention “tend to be especially useful for new products and features, or those undergoing redesigns.” It also warns that “what counts as ‘using’ a product can vary,” and notes that in some cases you might count someone as having adopted only once they have “successfully completed a key task.”
How do you calculate feature adoption rate?
Divide the users who completed the feature’s key action in a period by the active users who had access to it in the same period. Both halves of that fraction need a deliberate definition.
For the top, pick the action that means the feature did its job: an export that was scheduled, a report that was shared, an integration that synced data. Opening the page is not enough, because people open things to see what they are. For the bottom, count only users who could have used it: active in the period, on a plan that includes it, with a role that can reach it. Dividing by all signups makes every feature look unpopular.
A worked example: 2,000 active users are on plans with scheduled exports, and 300 of them schedule at least one export in the 30 days after launch. That is a 15% first-month adoption rate, which counts everyone who tried it; pair it with the retention row of the table to see how many of the 300 kept going. Run the same calculation each month and split it by plan or company size, and the trend tells you far more than the first figure.
What is a good feature adoption rate?
Whatever matches the audience the feature was built for. A change to the main workflow should reach most active users; a setting for account admins might be a success at a small fraction of users if most admins use it.
Benchmarks borrowed from other products rarely transfer, since they depend on how features are defined and counted. Set the target before launch instead: who the feature is for, what share of them you expect to use it within a month, and what retention after a few weeks would mean it worked. Then compare against that, and against the rate for similar features you have shipped before.
Why do users not adopt new features?
They did not hear about it, they could not find it when they needed it, or it did not help with anything they were trying to do. The fixes for each are different, so diagnose before acting.
The data usually points to the cause. If few users ever reach the feature, awareness or placement is the problem. If many reach it and few complete the key action, the feature itself or its first-use experience is the problem. If people complete it once and never return, it probably solved a problem they do not have often, or solved it worse than their old workaround. Asking a few of the users who tried and stopped is the fastest way to tell the last two apart.
How do you get more users to adopt a new feature?
Tell people in the places they already look, start with the users who asked for it, and put the feature where the need arises. Most of the work happens around the launch, not inside the feature.
- Announce it where users work. An in-app announcement reaches people at the moment they can try the feature. New feature announcement covers which channels suit which changes, and Changeloop’s widget shows the changelog inside your product next to a roadmap and a feedback box.
- Start with the people who asked. Users who requested a feature are the likeliest to adopt it, and telling them first gives you early usage to learn from. Changeloop handles this for feedback sent through its widget: once the pull request that closes the resulting GitHub issue merges and its changelog entry is approved, the issue gets a “Shipped” comment and the submitter sees the entry in the widget. A voice of customer program is what keeps track of who asked.
- Release in stages. Turning a feature on for a small group first lets you measure adoption and fix the first-use experience before everyone sees it. Feature flag release notes covers what to tell users at each stage.
- Put it in the path. Link the feature from the screen where the need comes up, from empty states and from related settings, so people meet it when it is useful rather than only on launch day.
- Show how in one step. A short example in the announcement, or a link to the exact page, removes the “where is it?” step that loses many users.
How do release notes affect adoption?
They decide whether people understand what they can do now. An entry that names the feature but not the task it helps with gets read and forgotten.
Write the entry around the user’s job: what they can now do, where to find it, and who it is for. “You can now schedule exports to run monthly, from Reports, Export, Schedule” gives a reader everything they need to try it. How to write release notes covers the structure, and the release notes template gives you a starting format.
When should you remove a feature that nobody uses?
After you have ruled out awareness and placement, and given it enough time to show a trend. Features with low adoption are often simply hard to find; removing one before checking that throws away the work for no reason.
If a feature still has low adoption and low retention after people demonstrably know about it, removal is often the right call, since every feature carries maintenance and support cost. Announce it in advance, tell the users who do rely on it what to use instead, and record the removal in the changelog like any other change. The notice period and the removal day are covered step by step in how to sunset a feature.
FAQ
What is the difference between feature adoption and product adoption? Product adoption measures new users starting to use the product as a whole. Feature adoption measures existing users starting to use one part of it. The HEART framework uses the same Adoption category for both and lets each team choose the specific metric.
How long after launch should you measure feature adoption? Measure from launch day, but judge it after a few weeks, once the announcement spike has faded. For a feature people need weekly, a month of data shows whether it stuck; for one used monthly or quarterly, wait for at least two cycles.
Should someone who used a feature once count as adopted? Count them in the adoption rate, which measures reach, and track retention separately to see who kept using it. A high trial rate with low retention means the feature is being found but is not helping, which calls for a different fix than low trial.
Does a changelog entry help adoption? It helps with awareness, the first of the three causes of low adoption. An entry that says what users can now do and where to find it reaches people who read the changelog, and an in-app widget reaches the ones who never visit it.
The technical claims in this article have not been independently reviewed. If something here is wrong, tell us and we will correct it.