How many profiles your product reaches. Rolling daily, weekly, and monthly active counts for every calendar day, the stickiness between them, and an honest freshness readout for the one page that is precalculated.
What an active profile is
A profile is active on a day when it had at least one session that calendar day. Everything on this page builds on that single definition.
The three counts are windows ending on each day. Daily active is that day alone. Weekly active is the rolling 7 days ending on the day, monthly active the rolling 30. Because the windows roll, the weekly and monthly lines move every single day, not once a week or once a month; a quiet weekend leaves the weekly line the moment it slides out of the window, seven days later.
The headline cards read the newest day in your selected range, so they are current snapshots, not period averages. The chart and the table below carry the same three windows for every day in the range, which is where trends live.
How fresh these numbers are
Active Profiles is the one page whose numbers are precalculated rather than queried live, because rolling windows over every profile are expensive to compute on each view. That earns it an honest freshness readout.
The age under the page title states when today's value was computed. Today recomputes when viewed, once its value is older than 5 minutes, so this age is usually small. Earlier days are re-verified by a background pass every hour for the trailing week, and a nightly pass reconciles the older history. When a range does not include today, the line shows when those days were last verified instead.
An amber strip appears only when something needs attention. Either the server is recomputing in the background and some days still read as zeros, or the background passes are overdue, which means the job server has stopped. The overdue strip offers Recalculate, which clears the cached history after a confirmation; the days you are viewing recompute right away and older days rebuild over the following nights.
Use Cases
See whether reach is growing
The growth question is a trend question, so the chart is the answer.
- Pick a range long enough for the story, at least a few weeks.
- Read the three lines together. Daily active is noisy by nature; weekly and monthly smooth it and show the direction.
- Turn on the date picker's comparison to lay the previous period underneath. The table gains per-day deltas, so you can see exactly which days carried the change.
- Growth in monthly active with flat daily active means you are adding profiles who visit rarely. Both rising together is the healthy shape.
Sessions asks the same question in visits; this page asks it in profiles reached. Read both when volume and reach disagree.
Judge stickiness
Stickiness is the share of your monthly audience you see on a typical day, daily active divided by monthly active.
- Read the stickiness card next to the other three. It contextualizes them; two products with the same monthly count can have wildly different daily pull.
- Above roughly 20 percent is healthy for consumer products. Utilities and seasonal tools sit naturally lower; team products used every workday sit far higher.
- Watch its trend rather than the single number. Stickiness falling while monthly active grows means acquisition is outpacing habit, the classic leaky bucket.
Check a suspicious dip
Before treating a dip as churn, rule out the boring explanations.
- Read the age under the title. If the numbers are hours old, the dip may simply be data that has not been recomputed yet.
- If an amber strip says the background passes are behind, the job server is not running; earlier days stop being re-verified, though today still recomputes when viewed. Recalculate rebuilds the range you are viewing right away.
- Check the Sessions section for the same days. If sessions also dropped, the dip is real behavior; if sessions look normal, suspect tracking changes, an SDK release, or a changed session cadence.
- A dip only in weekly or monthly active with normal daily active usually is not a dip at all, just a busy period sliding out of the rolling window.
FAQs
Why did weekly active change when yesterday was quiet?
The windows roll. Each day's weekly value covers the 7 days ending then, so days entering and leaving the window both move it.
Why do these counts differ from unique sessions in the Sessions section?
Sessions counts uniques over the whole selected period; this page counts them per day and per rolling window. A profile active on three days is one unique for the period but counts into three daily values.
Is today's number complete?
No, today is still accumulating and recomputes as you view it. It is final after midnight in the project's timezone.
What exactly does Recalculate do?
It clears the project's cached history after a confirmation. The range you are viewing recomputes from raw events immediately; older days rebuild when viewed or through the nightly pass. Nothing about the raw data changes.
Can stickiness exceed 100 percent?
No. Everyone active today is by definition active this month, so daily active is always at most monthly active.