Feature adoption rate calculator

Feature adoption describes how far a feature has become part of people’s work. One number cannot capture it, so product teams look at four dimensions:

The adoption rate is a share measured on a sample, so it has a margin of error. The calculator uses the Wilson interval, which stays correct for small numbers and for shares close to 0 or 100%. With 1,150 of 4,800 the 95% interval is 22.8% to 25.2%; with 23 of 96 (the same share) it is already 16.5% to 33.4%.

Repeat usage is the share of feature users who used it on at least two different days. High breadth with low repeat usage means people tried the feature and dropped it: there was interest, but not enough value.

Time to adopt

How long it takes from sign-up (or the feature release) to first use. The calculator takes the median and 75th percentile over users who have already used the feature, and separately the share of all users in the export who started within 7 and 30 days. The mean is misleading here: a few people who discover the feature two months later pull it far up.

There is no universal benchmark: it depends on who the feature is for. An accounting export matters to a tenth of customers, while creating a project matters to everyone. The scale in the calculator is therefore a set of rules of thumb for a feature meant for every active user, not an industry standard:

• under 10% — people barely find the feature, or it solves a rare task;
• 10–30% — normal for an advanced or niche feature;
• 30–60% — a sizeable part of the audience uses it;
• 60% and above — the feature is part of the core workflow.

The right denominator matters more than the cut-offs. If the feature exists only on a paid plan, divide by paying accounts; if only admins need it, divide by admins. Pendo’s 2019 Feature Adoption Report found that 80% of features in the average software product are rarely or never used. Most often that’s not because nobody needs them — it’s because users don’t know they exist. So low adoption is first a discoverability problem, and only then a question of value.

Compare a feature with itself over time and with similar features in your own product. If adoption grows after an interface change, confirm it with an A/B test — the A/B test calculator gives you the sample size.

When a feature is useful but rarely used, the problem is usually not the feature itself but the path to it:

• People cannot see it. It hides in a menu or appears only in a certain state. A tooltip next to the place where the feature is needed beats a release email.
• The value is unclear. The name says what the feature does, not why it matters. A short tour with an example of the result answers “what do I get”.
• Wrong people, wrong moment. Suggesting reports to someone with no data yet is pointless. Show the tooltip to users who completed the previous step, and hide it from those who already use the feature.
• The first run is hard. If a result needs five settings first, few people make it. Find where they drop off with the funnel calculator and shorten the path with the time-to-value calculator.

Measure the effect by cohort: adoption among users who saw the tooltip versus those who did not, and repeat usage after 2–4 weeks. The retention calculator shows how the feature relates to retention.

Common mistakes

• Events instead of users in the numerator: one active person generates hundreds of events.
• Internal and test accounts in the data inflate adoption of new features.
• The launch spike. In the first days people open a feature out of curiosity; the steady level shows up after a few weeks.
• A period that is too short for features needed weekly or monthly.

1. Pendo. The 2019 Feature Adoption Report — the share of rarely used features in software products.
2. Croll A., Yoskovitz B. Lean Analytics: Use Data to Build a Better Startup Faster. O'Reilly Media, 2013 — engagement metrics and choosing the denominator.
3. Wilson E. B. Probable Inference, the Law of Succession, and Statistical Inference. Journal of the American Statistical Association, 22(158), 1927 — interval for a proportion.
4. Hyndman R. J., Fan Y. Sample Quantiles in Statistical Packages. The American Statistician, 50(4), 1996 — percentile definition (type 7).