Skip to main content
Card class: HeroCategory: Retention

At a glance

Cohort Retention D7 % is the share of a signup or first-action cohort that returns and performs a qualifying event seven days after joining. It is the earliest strong retention signal you can read, much faster to act on than D30, and it tells you whether your onboarding and first-week experience actually earned a second visit. In Mixpanel terms, this is a retention report grouped by acquisition cohort, read at the day-7 column. Because it moves quickly, a fall in D7 is one of the first symptoms of a broken onboarding flow, a bad campaign cohort, or a product change that hurt the first-week experience. Watching D7 lets you correct course weeks before the damage shows up in D30 or in revenue.
What it countsThe percentage of users in a defined cohort (typically grouped by first event or signup date) who fired a qualifying return event on day 7 after joining.
Sample typeBackend API data from Mixpanel retention reports, computed per cohort against the project’s event stream.
Why it mattersD7 is the fastest reliable read on whether new users stick. It surfaces onboarding and first-week problems while you can still fix them for the next cohort.
Reading the valueA higher percentage is better. A sudden fall versus the recent baseline is the headline signal, treat a drop of more than a few points as worth investigating.
Currencypercent
Time window30D vsP
Alert triggerdrop >5pp vs baseline
Sentiment keymix_cohort_retention_d7
Rolesowner, marketing

Calculation

Vortex IQ reads your Mixpanel retention report grouped by acquisition cohort and takes the day-7 retention column. For each cohort, the numerator is the number of users who fired the qualifying return event on day 7 after their first event, and the denominator is the original cohort size. The card surfaces the blended D7 percentage across recent cohorts over the selected window and compares it against the recent baseline. The alert latches when the current reading falls more than five percentage points below that baseline. Because D7 only needs a cohort to age seven days, it updates far sooner than longer-horizon retention numbers.

Worked example

A representative reading of Cohort Retention D7 % for a typical merchant on Mixpanel. Suppose 800 shoppers signed up or placed a first order on 14 Jun 26. Seven days later, on 21 Jun 26, you check how many returned and fired a qualifying event such as a repeat session. If 312 came back, your D7 reads 39%. Your recent baseline across the last few cohorts was around 40%, so this cohort is healthy and no alert fires. Now imagine the next cohort reads 32% after a checkout redesign shipped, that is an 8pp fall against baseline, so the alert latches. You open the retention report, confirm the dip is concentrated in users who hit the new checkout, and roll back or patch the flow before the following cohort is affected. For deeper investigation, use Vortex Mind to trace upstream causes; for natural-language exploration, ask Ask Viq.

Sibling cards merchants should reference together

CardWhy merchants reach for it
mix_cohort_retention_d30The longer-horizon checkpoint. Compare D7 to D30 to see whether early retention holds.
mix_retention_curveShows the full decay curve so you can see how steeply users fall off after week one.
mix_worst_decaying_cohortPinpoints which acquisition cohort is losing users fastest, often a specific campaign.
mix_stickinessDAU/MAU stickiness is the habit signal that drives healthy week-one retention.
mix_new_usersNew vs returning users frames how much of your base is first-time versus repeat.

Reconciling against Mixpanel

Where to look in Mixpanel’s own dashboard: Open your saved Retention report, set the birth event to the same first action the card uses and the return event to the same qualifying event, then read the day-7 column for the cohorts in your window. The blended figure on the card should track the average of those cohort rows. Confirm the retention type (recurring vs unbounded) matches, because the two compute the day-7 column differently. Why the Vortex IQ value may legitimately differ:
ReasonDirectionWhat to do
Cohort maturity. Cohorts younger than seven days cannot yet report a D7 value.Vortex IQ excludes immature cohorts from the blendCompare only cohorts that have aged seven full days in both views.
Retention type. Recurring vs unbounded retention count the day-7 return differently.VariableMatch the retention type used by the card.
Event definition. The birth or return event differs between the report and the card.VariableAlign both event definitions before comparing.
Cross-connector reconciliation: check the D7 trend against second-session and repeat-visit signals in your ecommerce platform to confirm early returners are genuine shoppers. For divergence investigations, use Vortex Mind.

Known limitations / merchant FAQs

Q: How often does Cohort Retention D7 % update? It refreshes on the card’s regular cycle, and because D7 only needs a cohort to age seven days, the latest cohorts become measurable quickly. That speed is exactly why D7 is the preferred early-warning retention metric. Q: The alert fired but my D30 still looks fine. Which do I trust? Trust D7 as the early signal and D30 as confirmation. D30 will not reflect a recent change yet, because those cohorts have not aged a month. If D7 is falling, expect D30 to follow unless you intervene. Q: Why does D7 swing between cohorts? Small cohorts produce noisy rates, since a few returning users move the percentage. Watch the trend across several cohorts and weight larger cohorts more heavily before treating a single week as a problem. Q: Can I customise the alert threshold? Yes, the 5pp drop threshold and the baseline window are configurable per profile in the Sensitivity tab. Tighten it for high-volume stores where small movements are meaningful, or loosen it for low-traffic stores whose cohorts are naturally noisy.

Tracked live in Vortex IQ Nerve Centre

Cohort Retention D7 % is one of hundreds of KPI pulses Vortex IQ tracks across Mixpanel and 70+ other ecommerce connectors. Nerve Centre runs the detection layer; Vortex Mind investigates the cause when something moves; Ask Viq lets you interrogate any number in plain English. Start for free or book a demo to see this metric running on your own data.