How to measure feature adoption in a SaaS product

By Hasan · November 5, 2026 · 4 min read

How to measure feature adoption in a SaaS product

To measure feature adoption, send an event each time someone uses the feature, and tie those events to an account with group(). JustOneCX then shows, for each event name, the share of active accounts that used it in the period you choose.

The definitions here come from the analytics definitions guide.

What does feature adoption mean here?

For each event name, feature adoption is the number of accounts that sent that event in the period, divided by the number of accounts that sent any event in the period.

So if 40 accounts were active and 10 of them used your export feature, adoption of that event is 10 out of 40. The numbers here are only an illustration of the formula. They are not real data.

Friction events ($rage_click, $dead_click, $js_error) are left out of this report.

What do you need to send?

Feature adoption needs events tied to an account. The docs list the requirement plainly: use group() or an account in identify(), then track().

You want You must send
Feature adoption Events tied to an account: group() or an account in identify(), then track()

If a chart is empty, the page tells you which of these is missing.

How do you send the account?

Call group() after init() to set the account (company, team, organisation) the user belongs to:

group("acme-inc", { plan: "growth", seats: 12 });

The signature is group(accountId: string, traits?: Traits, options?: { accountHash?: string }): Promise<void>. Traits merge, so call it on every page load to keep them fresh.

You should also identify the person:

identify("user-42", { role: "admin", plan: "growth" });

See identify users with the SDK for more. Full reference: identify, group, track.

How do you send the feature event?

Call track() when the person uses the feature. It is tied to whatever identify and group calls came before it.

track("clicked_upgrade_cta", { plan: "growth" });

Use clear, stable names. For a feature, something like exported_report tells you more than click_7. The name is your choice. Properties can be strings, numbers, booleans or null.

If the action happens on your server, send the event from the backend with your Secret key and an accountId. The endpoint is POST /ingest/events/track:

curl -X POST https://api.justonecx.com/ingest/events/track \
  -H "Authorization: Bearer $JUSTONECX_SECRET_KEY" \
  -H "Content-Type: application/json" \
  -d '{"userId": "user-123", "eventName": "subscription_upgraded", "accountId": "acme", "properties": {"plan": "pro"}}'

Server events count toward analytics, segments and funnels like browser events. See server-side event tracking for SaaS.

What does the Analytics page show?

The Analytics page has several views. According to the docs they are:

  • Activation. Activation rate and median time to value for new identified users.
  • Feature adoption. Accounts that used each event, as described above.
  • Stickiness. DAU, WAU and MAU, plus DAU/WAU and WAU/MAU.
  • Friction signals. Rage clicks, dead clicks and JavaScript errors, sent by the SDK itself. No session recording.
  • Funnels. Two to four steps. See product funnel analysis for SaaS.

The overview is cached for up to 60 seconds, and the page shows "Updated N s ago".

How do you read the adoption numbers?

Read each event as a share of active accounts. A few things to check.

  1. Pick the period first. Adoption depends on the period you choose, so compare like with like.
  2. Compare features with each other. A low share is not bad on its own if the feature is for a small group. It is a signal to look closer when you expected wide use.
  3. Check the denominator. Accounts that sent any event in the period form the base. If few accounts are active, the share can swing a lot.
  4. Cross-check with stickiness. DAU, WAU and MAU show whether people come back.

These reading tips are general analytics practice. After you find a gap, you can act on it. For example, show a tour or an announcement to accounts that never used the feature. A segment with the Never performed event attribute fits this. See user segmentation for SaaS.

Why must adoption be account-based?

Adoption counts accounts, not single users. That fits B2B products, where one company has many users. An account with ten users who all use a feature counts once. This is why group() or an account in identify() is required.

Frequently asked questions

What do I need to send for feature adoption?

Events tied to an account. Use group() or an account in identify(), then track() the feature event.

How is feature adoption calculated?

For each event name: accounts that sent it in the period divided by accounts that sent any event in the period. Friction events are left out.

Why is my adoption chart empty?

An empty chart says which data is missing. Check that your events are tied to an account: call group() (or pass an account in identify()) before track().

Can I send adoption events from my server?

Yes. Use POST /ingest/events/track with the Secret key, a userId and an accountId.

See how it works in the live demo.

See it working in 2 minutes

Try the live demo

Related