Brett LindenbergUser engagement metrics measure how often people come back to a product, how much of it they use, and whether they still use it weeks later. The core set is active users (daily, weekly, and monthly), stickiness (DAU/MAU), cohort retention, sessions per user, feature adoption, and completion of the main task the product exists for. Before you choose from that list, define what counts as "active" for your product and how often a healthy user should return. A language app expects a learner every day, a reporting tool once a week, and a compliance course once a year, so each one needs different metrics. In the short demo below, you see one drop-off in a training lesson and choose what to inspect first.
Which metrics measure user engagement?
Each metric answers a different question. Report rates and averages per user, so that growth in the number of users does not hide a drop in use.
| Metric | How to calculate it | What it tells you | Watch out for |
|---|---|---|---|
| Active users (DAU, WAU, MAU) | Distinct users who did a meaningful action in the last 1, 7, or 30 days | How many people use the product in each period | An app open counts as "active" unless you define the action |
| Stickiness (DAU/MAU) | Daily active users ÷ monthly active users × 100 | How many monthly users return on a typical day | Fits daily products; use WAU/MAU for weekly ones |
| Cohort retention | Users from a start cohort still active in week or month N ÷ cohort size × 100 | Whether people keep using the product | Compare cohorts of the same age |
| Churn rate | Users or accounts lost in a period ÷ users or accounts at the start × 100 | How fast you lose users | Decide how long a user must be inactive to count as lost |
| Sessions per user | Sessions ÷ active users | How often people return | Confused users also open many sessions |
| Engagement time per user | Time the product was in the foreground ÷ active users | How much people use it in each period | In a support or self-serve tool, more time can mean more trouble |
| Feature adoption rate | Users who used a feature ÷ active users × 100 | Whether a feature found its audience | Count repeat use, since a first click can be curiosity |
| Activation rate | New users who reach a first-value action within a set number of days ÷ new users × 100 | Whether onboarding works | Pick an action that predicts retention |
| Task completion rate | Users who complete a key task ÷ users who start it × 100 | Whether people succeed at the main job | Pair it with time to complete and error rate |
| Satisfaction | A survey score, such as a rating after a task | How users feel about the product | Ask at the same moment each time |
Google Analytics 4 includes some of these as standard metrics. The Google Analytics Data API reference defines DAU / MAU as the rolling share of 30-day active users who were also active in the last day, and it offers DAU / WAU and WAU / MAU in the same form. It counts user engagement as the time your site or app was in the foreground of the user's device. In GA4, an engaged session lasts longer than 10 seconds, has a key event, or has two or more page or screen views.
Choose metrics with the HEART framework
Google researchers Kerry Rodden, Hilary Hutchinson, and Xin Fu published the HEART framework at the CHI 2010 conference. HEART stands for Happiness, Engagement, Adoption, Retention, and Task success. The authors pair it with a Goals-Signals-Metrics process: write the goal, name the user behavior that would show progress toward it, and then build the metric from that behavior.
The paper gives two rules that still hold. Report engagement as an average per user, because a higher total can come from more users with no more use. And fit the categories to the product: the authors note that engagement may mean little when people must use a product for their work, and that such teams can focus on happiness or task success. Their Gmail example shows how specific a good metric can be: the team tracked the share of active users who visited on five or more days in the last week, and found that it strongly predicted longer-term retention.
How to measure user engagement in 6 steps
- Write the goal in one sentence. For example: "New team admins build their first report in their first week."
- Define an active user by a meaningful action. Opening the app is a weak signal. Creating, sending, completing, or answering something is a stronger one.
- Set the natural frequency. If a healthy user returns daily, track DAU and DAU/MAU. If weekly, track WAU and WAU/MAU. If people use the product once per project, track cohort retention by month.
- Choose one primary metric and one guardrail metric. The guardrail warns you when a change raises the primary metric and hurts something else, such as support tickets or satisfaction.
- Track events, then group users into cohorts by the week or month they started. Split each cohort by plan, role, device, and acquisition channel.
- Record a baseline, change one thing, and compare the next cohorts with the earlier ones at the same age.
Example: how Duolingo measures engagement
Duolingo, a language-learning app, reports its engagement metrics every quarter. In its Q2 2026 shareholder letter, filed with the SEC on August 5, 2026, the company says it evaluates engagement mainly with daily active users and uses monthly active users as a supplementary metric. It reported 58.7 million DAUs and 140.6 million MAUs for the quarter, which is a DAU/MAU ratio of about 42% by our calculation.
Its main stickiness metric is Current User Retention Rate (CURR). A "current user" is a daily active user who was also active at another time in the previous seven days, and CURR is the share of those users who come back the next day. Duolingo calls CURR one of its most important leading indicators, because small gains compound into large DAU gains, and it reported CURR at an all-time high of 84%.
What is a good DAU/MAU ratio?
No published threshold fits every product, so set your target from your own cohorts. The ratio depends on how often a healthy user should return. If every user of a payroll tool logs in on exactly two days a month, its DAU/MAU is about 7%, and those customers can still be satisfied. A daily app such as Duolingo sits near 42%. For a product used weekly, compare WAU/MAU. For a product used per project, use cohort retention.
How to find user engagement metrics
Most teams need three or four sources. Start with the question, then go to the source that records it.
| Question | Where to look |
|---|---|
| How many people use the website or app, and for how long? | GA4: active users, engaged sessions, engagement time, sessions per user, and key events |
| Do new users keep coming back? | An event-based product analytics tool, or a cohort query on your own database |
| Which features do people use? | Feature events in product analytics, or usage tables in your database |
| Do customer accounts use what they pay for? | Your database: active seats ÷ paid seats, and last-active date per account |
| Do learners finish and pass? | LMS reports for completion and score; the content tool for question-level detail |
| Where do viewers stop watching a video, and what do they answer? | Your video platform's retention and per-viewer reports |
| How do users feel? | A short in-product survey at a fixed moment |
For B2B products, account-level data often matters more than user counts. A customer with 50 paid seats and 8 active users is at risk even if those 8 users log in every day. For the methods behind event and path analysis, see our guide to behavioral analytics.
LMS user engagement: what your LMS can and cannot see
An LMS records enrollments, launches, completions, scores, and time for each learner. For content packaged as SCORM, that data comes from a fixed data model. In SCORM 1.2, a course reports cmi.core.lesson_status, cmi.core.score.raw, and session time. SCORM 2004 splits the status into cmi.completion_status and cmi.success_status (Rustici Software's SCORM run-time reference). Both versions also define an interactions array for individual responses. The LMS sees only what the course sends, so a package that sends status and score tells the LMS that a learner finished and how they scored. With only those two values, the LMS cannot show where a learner lost interest or which wrong answer they chose.
Track these metrics for a learning platform:
- Start rate: learners who launched a course ÷ learners enrolled
- Completion rate by course and by module
- Drop-off point inside a module, which needs data from the content itself
- First-attempt correct answers, per question and per assessment
- Time to complete, compared with the time you planned
- For optional learning: active learners per week, and learners who start a second course
- A short satisfaction rating at the end of each module
For mandatory training, the HEART point about required use applies. Learners take a compliance course because they must, so logins and completions say little about interest. Judge those programs by first-attempt correct answers and by behavior on the job. Our guide to measuring training effectiveness covers the full method, and our guide to measuring student engagement covers classroom and online courses.
Mindstamp is interactive video software that works alongside your LMS. A Mindstamp SCORM package, in SCORM 1.2, SCORM 2004 3rd edition, or SCORM 2004 4th edition, sends completion and a score to the LMS on supported plans. Each answer, click, and path stays in Mindstamp reports for each learner. See LMS compatibility for the ways to deliver a video in an LMS and what each one records.
What engagement metrics matter for learning apps?
For a consumer learning app, track daily return, practice volume, and learning outcomes together. Useful measures are DAU, the next-day return rate of regular learners (Duolingo's CURR is one version), lessons or practice minutes per active learner per week, and scores on checks of what the learner knows. A language app, for example, might track minutes of speaking practice per learner each week. Duolingo's Q2 2026 letter names user retention and learning outcomes as its priorities for 2026, ahead of monetization.
How do you measure the engagement and effectiveness of training videos?
Measure engagement with how much of the video each learner watched and where they stopped. Measure effectiveness with answers to questions placed at the moment each concept is taught, and later with behavior on the job. Our guide to engagement rate shows how to calculate the four numbers for a training video: watch rate, drop-off point, participation rate, and first-attempt correct rate.
Read these numbers together, as in the demo above. A sharp drop at 04:20, just before a question that 42% of viewers answer correctly, points first to the explanation before 04:20. The final completion rate cannot tell you whether the explanation or the question caused the trouble, and a low correct rate can mean a real knowledge gap or a flawed question. In Mindstamp, a question pauses the video until the learner answers. Watched percentage counts only the unique seconds a learner played (how Mindstamp reports work), and each video's Reporting tab has a drop-off graph. For metric definitions and watch-rate figures by video length, see our guide to video engagement metrics.
How do you use video engagement data to improve training content?
- Find the steepest drop in furthest point, the latest moment each learner reached, across views (furthest point and last point).
- Open the answer breakdown for the next question. The wrong answer most learners chose usually names the misunderstanding.
- Rewrite or re-record the explanation that comes before it.
- Add an answer action that sends a learner who answers wrong back to the explanation (retries and answer actions).
- Compare correctness and drop-off for the next group of learners with the last one.
Our overview of video analytics for training covers this process in more depth. To turn the gains into a financial case, see how to calculate training ROI.
How do you measure user engagement in a self-guided demo besides completion rate?
Completion tells you that a viewer reached the end. A self-guided demo also records the choices a buyer makes on the way, and those choices are often more useful to sales than completion.
- Start rate: viewers who started the demo ÷ visitors to the page
- Step reach: the share of viewers who reached each step, and the step with the largest loss
- Path choice: which use case, role, or feature each viewer chose to see
- Call-to-action rate: clicks on "book a meeting" or "start a trial" ÷ viewers who reached the button
- Repeat views by the same viewer, and new viewers who arrive from a forwarded link
- Identified viewer rate: the share of viewers sales can follow up with
A demo built with branching in Mindstamp records the path each viewer chose, with each click and its timecode. Lead capture or a personalized link attaches the session to a person. On supported plans, the HubSpot integration adds view and answer events to the contact record after the session ends. See interactive product tours for examples.
Which viewer analytics show engagement drop-offs during a live workshop stream?
Watch concurrent viewers over time, and read the curve against your agenda. On YouTube, the Live Control Room shows concurrent viewers, chat rate, and average view duration during the stream. Chat rate is the number of live chat messages per minute. After the stream, YouTube Studio adds average and peak concurrent viewers and a key moments for audience retention report.
- Write down the start time of each agenda segment before you go live.
- After the stream, mark those times on the concurrent viewer curve. A segment where the count falls steadily is your drop-off.
- Compare chat messages per minute across segments. A quiet chat during a hands-on segment often means people are stuck.
- For each poll, divide responses by concurrent viewers at the moment of the poll.
Mindstamp does not stream live. Once the recording is available, you can add questions and chapters to it and measure the on-demand audience viewer by viewer. Our guide to RTMP live streams and the webinar repurposing use case cover that handoff.
Mistakes that distort user engagement metrics
- Counting logins or app opens as engagement. Define active by an action that delivers value.
- Reporting totals. More users can raise total sessions while use per user falls.
- Treating more time as better in every product. In a support tool, success may mean less time.
- Mixing new and long-time users in one retention figure. Use cohorts.
- Leaving your own team's test sessions in the data. Exclude internal users and test accounts.
- Reading required use as interest. For mandatory tools and courses, measure task success.
- Changing several things at once, so you cannot tell which change moved the number.
Common questions about user engagement metrics
What is the difference between user engagement and user retention?
Engagement describes how often and how much people use a product within a period. Retention describes whether they are still using it in a later period. Some engagement metrics predict retention. In the HEART paper, the Gmail team's measure of users active on five or more days a week did, so you can use a metric like that as an early warning.
How is user engagement different from content engagement?
User engagement measures how people use a product over time. Content engagement measures what people do with one article, email, post, or video. See our guides to content engagement metrics and video engagement metrics for those, and our guide to measuring content marketing ROI to connect content to revenue.
Can engagement metrics replace formal assessments?
No. Correct answers to questions inside a lesson show whether learners understood each concept as they learned it, and they help you find struggling learners early. Certification and high-stakes compliance still need an assessment with controlled conditions. Use engagement data to decide who needs support before that assessment.
How many engagement metrics should a team track?
Track one primary metric for the goal you are working on, one guardrail metric, and a few supporting metrics you check when the primary one moves. A dashboard with many equal metrics makes it easy to pick whichever number supports the answer you wanted.
To start, write down what an active user does in your product and how often a healthy user does it. Then choose the one metric from the first table that matches that frequency, record this month's baseline by cohort, and check it again after your next release.
