A cohort table follows groups that share a start condition—such as first visit, signup, first lead, or first purchase—and shows how many satisfy a return condition over later periods. It helps answer whether newer groups are retaining better, where engagement drops, and whether a campaign acquired people who continue to receive value. The table is only meaningful when inclusion, return, time granularity, and calculation type are explicit.
VISUAL LESSON
What you will learn
- 01Define cohort inclusion, return criteria, granularity, and calculation type.
- 02Read a retention row across time and compare cohorts at equal maturity.
- 03Distinguish retention change from acquisition-volume change.
ILLUSTRATIVE WEEKLY COHORT
The first cohort shrinks, then begins to stabilize
GA4 COHORT MAP
Configure the table before interpreting color
Select first touch, an event, transaction, conversion, or another specific event and define daily, weekly, or monthly granularity.
Set the later event, transaction, conversion, or other condition and choose standard, rolling, or cumulative calculation.
Review equal-age cells, sample size, source or device breakdowns, data thresholds, and changes that occurred between cohorts.
THE RETENTION TABLE
Start together, return over time
THE LEAD ATLAS METHOD
Lead Atlas Data can produce a campaign-specific business-contact cohort for selected markets, locations, and categories, allowing the customer to compare response, qualification, and continued engagement for each researched batch without mixing it into other acquisition cohorts.See how custom list research works ↗Define inclusion and return
Inclusion determines who enters a row and when: first visit, signup, first purchase, qualified lead, or another event. Return determines what later behavior counts as retained value.
Choose events that match the business model. A page view may be too weak for a subscription service, while a repeat purchase may be too slow for an onboarding diagnostic.
Choose time and calculation type
Daily, weekly, and monthly granularity answer different questions. Match the window to the normal customer rhythm and allow newer cohorts enough time to mature before comparison.
Standard retention counts return in each period, rolling retention requires continued return through the period under the tool’s definition, and cumulative views count users who returned by a point. Label the type clearly.
Read across a row and down a column
Reading across one row shows how a single starting cohort changes as it ages. Reading down the same period column compares different cohorts at equal maturity, such as Week 2 for every acquisition week.
Do not compare Week 4 for an old cohort with Week 1 for a recent cohort and call the difference improvement.
Separate volume from retention quality
A large campaign can create more returning customers while producing a lower retention rate; a smaller campaign can show a higher rate with fewer retained people. Report both cohort size and percentage.
Break down source, campaign, plan, market, device, or activation status when evidence and privacy allow. Investigate mix changes before crediting one onboarding or product change.
Turn patterns into a customer hypothesis
A steep early drop may point to expectation mismatch, weak activation, technical failure, or a one-time use case. A later plateau may reveal a core group receiving recurring value. Qualitative research and operational data are needed to explain the pattern.
Exercise: define one inclusion event and one return event, build three weekly cohorts, compare the Week 2 column, and list the acquisition or experience changes that could plausibly explain the difference.
THE TAKEAWAY
Define cohort entry and return behavior first, read across rows and down periods, compare matched maturity, and connect retention changes to customer value rather than acquisition volume alone.OFFICIAL REFERENCES