The audience going quiet. Profiles counted by how long since their last activity, at five thresholds from a week to ninety days, with a preview of exactly who they are and one click to save them as a cohort to win back.
What slipping away means
Every request a profile's device sends stamps the profile with the time of that request. Slipping Away reads those stamps and counts how many profiles have gone quiet: last active more than 7, 14, 30, 60, and 90 days ago.
The five counts are cumulative on purpose. Someone dark for three months is inside all five, so each row answers one question, "how many have been away at least this long", and the 90-day row is always a subset of the 30-day row. The recency bar above the table undoes the overlap: it splits the same audience into non-overlapping bands, from active in the last week to dark for ninety days or more, so the shape of the whole audience is visible in one bar.
This page describes the audience as it is right now. There is no date range; the boundaries move with the clock, and a profile leaves the counts the moment it comes back.
From number to action
A churn number you cannot act on is trivia. This page is built so the path from the number to the action is short.
Every count in the threshold table opens the actual profiles behind it, the same profiles, filtered the same way. From there, one button saves them as a cohort, seeded with the same inactivity condition, and the cohort is where action lives: target it with a push or an in-app message, watch it in Drill, compare it in Cohorts.
The saved cohort keeps evaluating. As profiles come back, they stop matching the condition and leave it, so a re-engagement campaign aimed at the cohort automatically stops reaching the ones it already won back.
Use Cases
Size the at-risk audience
Know how much of the audience is at risk, and whether it is getting worse.
- Open Slipping Away. Read Total profiles first; every share on the page is against it.
- Read Slipping >30d and >90d. The 30-day figure is the actionable pool; the 90-day figure is closer to lost.
- Read the risk index. It compresses the whole distribution into one number of days, so it is comparable: across projects, and against what it was last month.
- Read Audience by recency for the shape. A healthy product front-loads the bar on the active side; a tall 90d+ band is accumulated loss.
- Come back on a rhythm. The page has no history, so the trend lives in your habit of reading it, or in saving the thresholds as cohorts and watching those counts over time.
Find who is slipping, and act
Turn a threshold into a re-engagement audience.
- Narrow the audience first if the campaign has a shape: the filter takes profile properties and cohort membership, so "paying profiles going quiet" is one chip away.
- Click the count on the threshold that matches your intent. Thirty days is the usual re-engagement point; seven days catches profiles while the habit is still warm.
- Read the preview. These are the actual profiles, with the filter and the inactivity boundary shown above the list; open one to sanity-check that the audience is who you think.
- Create a cohort. The cohort is seeded with the same condition, named after the threshold, and re-evaluates on its own.
- Aim your re-engagement at the cohort, a push, an in-app offer, an email export, and watch it shrink as profiles return.
Find where churn concentrates
Find out whether going quiet is evenly spread, or concentrated somewhere fixable.
- In the by-segment card, pick the threshold and a dimension: platform, app version, country, or any list-valued property, including cohort membership.
- Read the bars. Each is the slipping count inside one segment value, so a bar out of proportion with the segment's size is the finding.
- Switch the threshold. A segment that dominates at 7 days but not at 90 loses profiles briefly; one that dominates at 90 loses them for good.
- Cross-check with the filter: set the top-of-page filter to that segment and read the whole page for it, its own risk index, its own recency shape.
- Act where the concentration is: an old app version that deserves a forced update, a platform with a broken onboarding, a country with a payment problem.
FAQs
Why do the threshold rows add up to more than the audience?
They overlap by design. Every row contains all the rows below it; the recency bar shows the same audience without overlap.
Why is there no date picker?
The question is about now: who is currently quiet, and for how long. History comes from saving a threshold as a cohort and watching it.
Is a slipping profile a churned one?
Not yet. Slipping means quiet past a threshold; some of those profiles return on their own, more return when nudged. The 90-day row is the closest thing to churned this page shows.
Why did the numbers change between two visits today?
The boundaries are measured from the moment of reading, so profiles cross them continuously, in both directions.
How does this relate to Loyalty's active in 7 and 30 days?
They are close complements, not exact ones: Loyalty counts who had a session within a window, while this page counts whose last activity of any kind, a session or any other request, is older than a threshold. Loyalty groups by lifetime session count; this page groups by silence.
What happens to the cohort I saved as profiles come back?
It shrinks. The cohort keeps evaluating its condition, so a profile's next activity removes it, which is exactly what you want a re-engagement audience to do.