At a glance
Fastest-Decaying Cohort surfaces the single acquisition cohort that is losing users faster than any other in your Mixpanel retention data. Mixpanel groups users by the week they first appeared and tracks how many come back over time; this card scans those curves and lifts the worst one to the top. A cohort that collapses quickly is usually a story about that week, a paid campaign that bought low-intent traffic, an onboarding regression, or a promotion that pulled in one-time bargain hunters. By naming the worst week, the card tells you exactly where to look rather than leaving you to read every cohort curve by hand.
| What it counts | The acquisition cohort with the steepest retention decay across the lookback window, identified from Mixpanel’s cohort retention curves. The card names the cohort week and its retention at the alert checkpoint. |
| Sample type | Backend API data from Mixpanel retention reports, evaluated across all cohorts in the window to find the fastest decline. |
| Why it matters | Acquisition spend is wasted when a cohort churns out within days. Spotting the worst week early lets you trace it to a source or an onboarding change before the next campaign repeats the mistake. |
| Reading the value | The card shows the cohort week and its retention percentage at the checkpoint. A lower percentage means a steeper fall. Compare it to the typical curve to judge how unusual the week was. |
| Currency | percent |
| Time window | 90D |
| Alert trigger | any cohort retention <10% by D14 |
| Sentiment key | mix_worst_decaying_cohort |
| Roles | owner, marketing |
Calculation
Vortex IQ reads the cohort retention curves Mixpanel produces for the lookback window, where each cohort is the set of users who first triggered your activation event in a given week. For every cohort, it measures how retention falls across the standard checkpoints (for example D1, D7, D14, D30) and ranks the cohorts by how quickly they decay. The cohort with the steepest early loss is surfaced as the fastest-decaying one. The alert latches when any cohort’s retention sits below the configured floor at the configured checkpoint, by default under ten percent by day fourteen.Worked example
A representative reading of Fastest-Decaying Cohort for a typical merchant on Mixpanel. Suppose your weekly cohorts usually hold around 22% of new users by day fourteen. Looking across the last 90 days, most weeks cluster near that figure, but the cohort first seen in the week of 02 Jun 26 sits at just 7% by D14. The card surfaces that week as the fastest-decaying cohort. You cross-reference your campaign calendar and find that week ran a deep discount promotion that drew a wave of one-time buyers. The decay is not a product problem, it is a traffic-quality problem tied to that single promotion. You note it so the next discount is paired with a stronger onboarding follow-up. For deeper investigation, use Vortex Mind to trace upstream causes; for natural-language exploration, ask Ask Viq.Sibling cards merchants should reference together
| Card | Why merchants reach for it |
|---|---|
mix_retention_curve | The full set of cohort curves this card scans. Open it to see where the worst week sits against the rest. |
mix_cohort_retention_d7 | Early-life retention at day seven, the first checkpoint where a fast decay becomes visible. |
mix_cohort_retention_d30 | Longer-horizon retention; confirms whether the decay continues or stabilises after the early drop. |
mix_stickiness | Stickiness shows whether engagement across the whole base is falling, not just one cohort. |
mix_returning_users | New vs returning split puts the decaying cohort in the context of repeat-visit behaviour overall. |
Reconciling against Mixpanel
Where to look in Mixpanel’s own dashboard: Open the saved Retention report for your activation event and switch the view to weekly cohorts over the last 90 days. The retention grid shows each cohort week as a row and each checkpoint as a column. Find the row with the lowest value in the day-fourteen column; that should match the cohort the card names. Make sure the retention type (for example unbounded vs bounded) and the activation event match the ones the alert uses, or the curves will look different. Why the Vortex IQ value may legitimately differ:| Reason | Direction | What to do |
|---|---|---|
| Checkpoint definition. D14 may be measured as exactly day fourteen or as a rolling window. | Variable | Match the checkpoint basis used by the card before comparing. |
| Activation event. A different first event redefines who belongs to each cohort. | Variable | Confirm the same activation event is selected in both places. |
| Maturing cohorts. The most recent weeks have not finished their D14 window yet. | Vortex IQ may read lower for new cohorts | Exclude cohorts younger than the checkpoint, or wait for them to mature. |