Stickiness: Meaning, Formula, and Repeat Use
Stickiness measures how frequently users return to a product and repeat an activity. In product analytics, Amplitude defines stickiness as the number of different days or weeks on which a user performs an event within a longer period. DAU/MAU is a common way to summarize that frequency, but its meaning depends on how the active users and time windows are counted.
How to calculate stickiness
The DAU/MAU ratio used by Google Analytics divides daily active users by monthly active users:
Stickiness (%) = daily active users ÷ monthly active users × 100
DAU means daily active users, WAU means weekly active users, and MAU means monthly active users. These are counts of distinct users who meet an activity rule during the relevant period. Repeated actions by the same user within a day do not add more users to that day’s DAU.
Two calculations commonly carry the DAU/MAU label: a daily snapshot and a monthly average. State which calculation a report uses before interpreting its percentage.
Daily snapshot: who was active recently?
In Google Analytics’ stickiness documentation, the windows are:
| Metric | Active-user window |
|---|---|
| DAU | Last 24 hours |
| WAU | Last 7 days |
| MAU | Last 30 days |
Its DAU/MAU ratio reports the share of the last 30 days’ active users who were also active in the last 24 hours. The numerator’s users belong to the denominator’s population when the activity and identity rules match.
A snapshot describes activity at that point in time. A 100% snapshot means every user in the broader window was active in the shorter window. It does not establish that every user was active on every earlier day.
Monthly average: how frequently did monthly users participate?
For a completed calendar month, Mixpanel describes a calculation using average DAU:
- Define the activity that qualifies a user as active.
- Count distinct active users separately for every day of the month.
- Sum the daily counts and divide by the number of days in that month.
- Count distinct users active at least once during the entire month.
- Divide average DAU by that monthly unique-user count and multiply by 100.
Written as formulas:
Average DAU = sum of daily unique active-user counts ÷ days in the month
Monthly-average stickiness (%) = average DAU ÷ MAU × 100
MAU requires deduplicating users across the whole month. Summing daily counts produces active user-days, because the same user can appear on several dates. That sum is not MAU.
Include days with no qualifying activity when calculating the daily average. Averaging only days that contain events changes the denominator of the average and overstates frequency across the full month. Use a completed period so the numerator and denominator cover the same observation window.
If MAU is zero, the division is undefined. Report that no users qualified during the period rather than displaying a calculated percentage.
What monthly-average DAU/MAU actually means
The monthly-average formula can be rearranged into a more direct frequency measure. Let the month contain D days and M distinct active users. Let U be the total active user-days, counting each user at most once per date.
Average DAU is U ÷ D. Therefore:
Monthly-average stickiness as a fraction = U ÷ (D × M)
Across the monthly active population, U ÷ M is the average number of active days per user. It follows directly that:
Average active days per monthly user = stickiness as a fraction × days in the month
This is a mathematical interpretation of the fixed-period formula. It summarizes how much of the observation period the average monthly active user participated in. It does not identify the particular users who returned or the sequence of their visits.
The same calculation also gives its bounds. With at least one monthly active user, each user contributes at least one active day and at most D active days. Monthly-average stickiness therefore ranges from 1 ÷ D to 1, expressed as a fraction. The minimum changes with the length of the month.
At the upper bound of 100%, every monthly active user qualified on every day of the month. At the lower bound, every monthly active user qualified on only one day. A positive monthly-average ratio can therefore exist without any repeated use within the month.
This interpretation requires one fixed monthly population. Averaging daily rolling DAU/MAU percentages is a different calculation: each day’s MAU can contain a different set of users. That average cannot be converted directly into average active days for one fixed group.
Define active use and user identity
The formula is simple; the activity rule determines what it measures. Amplitude’s glossary defines an active user by default as someone who logs at least one event during the selected interval, with inactive events excluded. A platform’s default active-user count and a measure of a particular completed action can therefore answer different questions.
Choose the event or event set that represents the behavior being measured. Apply the same rule to the daily numerator and monthly denominator. A ratio based on different qualifying actions in the two windows no longer measures the frequency of one consistent behavior.
Passive events need explicit treatment. Amplitude’s activity-status documentation identifies notification-sent and message-received events as events that can occur without a user’s action. Marking an event inactive removes it from active-user metrics, and the change applies immediately and retroactively.
Record event-definition changes with the metric. Historical values recalculated under a new event rule are a new measurement of past activity; a change in those values alone does not establish a change in user behavior.
Identity also affects both counts. According to Amplitude’s unique-user documentation, a stable user ID allows activity on multiple devices to belong to one user. Creating a new user ID for an existing user instead makes that person a different unique user in Amplitude.
Use the same identity resolution in both windows, and label the unit being counted. Counts of devices, signed-in users, and organizations represent different populations. A user-based DAU/MAU ratio requires user-based counts throughout.
For reproducible reporting, record the qualifying events, identity unit, filters, reporting time zone, window boundaries, and aggregation method. Keep those settings consistent across comparisons.
Look at the distribution of active days
An average does not describe how frequency is distributed across users. Alongside DAU/MAU, report the number of active days per user over a single completed period.
Amplitude’s stickiness analysis documentation distinguishes two views:
- Exact frequency: the percentage of users who performed the event on exactly a specified number of days.
- Cumulative frequency: the percentage who performed it on at least a specified number of days.
The exact-frequency view shows how activity is spread across occasional and frequent use. The cumulative view answers how much of the active population meets a chosen frequency threshold. Specify the event, period, and threshold when presenting either view.
Amplitude also notes that a user can appear in different buckets across different weeks or months. For a straightforward distribution of one population, use one observation period rather than interpreting a multi-period chart as mutually exclusive groups of people.
Population composition matters mathematically. Monthly-average stickiness is the average active-day count divided by the period length. Adding users whose active-day counts are below the existing average lowers that average, even when existing users’ counts stay unchanged. Removing lower-frequency users raises it without requiring anyone else to become more active.
Report MAU alongside the ratio, and compare consistent groups separately. Distinguishing new users from established users helps separate changes in population composition from changes in repeat behavior. A ratio alone cannot make that distinction.
What is a good stickiness ratio?
There is no universal DAU/MAU target that applies across different activity definitions and expected use frequencies. Sequoia’s product-health guidance advises choosing an active-user metric according to how frequently the product is expected to be used and states that DAU/MAU depends strongly on the product and its expected usage.
Assess frequency against the recurrence of the activity being measured. Daily activity is a relevant lens for a daily use pattern. Less frequent activity needs an observation period long enough to capture its recurrence. A low daily ratio alone does not show that users have stopped returning when the activity is needed.
Before comparing a published benchmark, check its qualifying action, identity unit, population, window, and use of a snapshot or average. A percentage calculated under different rules cannot establish whether a particular product is performing well.
Google Analytics also provides DAU/WAU and WAU/MAU ratios. These compare shorter and longer activity windows; they still require an activity definition and a cadence appropriate to the question.
Stickiness, retention, and engagement measure different things
Stickiness describes frequency within an observation period. Retention follows return after a defined starting point. Amplitude’s retention documentation explains that retention analysis compares a starting event with a subsequent return event for the selected users.
DAU/MAU does not perform that cohort-return calculation. It does not establish next-day retention, next-month retention, or continued use after the measured month. Measuring those outcomes requires following the relevant users into the return period.
Frequency is also distinct from the amount of activity within an active day. Performing more actions on a day that already qualifies does not change that user’s contribution to DAU. Assess completion or depth of use separately when those outcomes matter.
For periodic activity, Amplitude’s same retention documentation describes usage interval analysis, which examines time between critical events. That provides another way to study recurrence when daily participation is not the appropriate expectation.
How to investigate and improve repeat use
Start with a stable measure of the activity that is meant to recur. Compare its daily counts, monthly population, and active-day distribution across equivalent completed periods. Review event and identity changes before interpreting movement in the ratio.
Then examine the path to that activity. Measure whether users reach it, complete it, and return for it at the expected interval. Investigate drop-off at the relevant step, and evaluate product changes against that specific behavior.
Assess changes with more than the headline percentage. More frequent completion by a consistent group, return in subsequent periods, and growth in the qualifying population answer different questions. Track them separately so an improvement in one does not conceal deterioration in another.
A report should state the stickiness formula and show its underlying active-user counts. Add the frequency distribution and retention measure that fit the activity’s cadence. Together, these make repeat use interpretable and provide a concrete basis for deciding what to improve.