Retention Report Overview
The retention report groups users into cohorts based on a specific event (for example, install) and shows, in matrix form, whether users came back (were retained) after a certain period (D1, D3, D7, D14, D30) following the install. By comparing retention rates across dimensions such as ad channel, campaign, and creative, you can identify which acquisition sources bring in users who stay the longest.
1. Cohort and Retention Concepts
| Term | Description |
|---|---|
| Cohort | A group of users who performed the same reference event at the same point in time. The default reference is install (install). |
| Retention | The percentage of users grouped into a cohort who became active again (returned) after a certain period. |
| D-day | Indicates how many days have passed since the cohort reference date. D1 means the next day, and D7 means 7 days later. |
| Cohort size | The number of unique users grouped into that cohort. It serves as the denominator for the retention rate calculation. |
2. Difference from the Attribution Report
Both the retention report and the attribution report group data by channel and campaign dimensions, but what they measure is different.
| Category | Attribution Report | Retention Report |
|---|---|---|
| Measurement perspective | Performance (acquisition, conversions, cost, etc.) | Retention (whether users return after install) |
| Key metrics | Installs, conversions, ROAS, etc. | D1/D3/D7/D14/D30 retention rates |
| User grouping | Based on attributed events | Based on cohort (reference event) |
| Purpose of use | "Which channel brought in the most" | "How long the acquired users stay" |
In summary, the attribution report shows how many users you brought in (performance), while the retention report shows how long the acquired users stay (retention). Looking at both reports together lets you distinguish between "channels that bring in many users but lose them quickly" and "channels that bring in fewer users but keep them longer."
3. Screen Layout
When you enter 리텐션 리포트 in the left menu, the screen is divided into a left filter panel and a right results area.
| Section | Description |
|---|---|
| Left filter panel | Configures the cohort period, cohort reference event, return event, grouping dimension, retention days, filters, platform, and minimum cohort size. At the top there are 초기화 and 쿼리 실행 buttons. |
| Weighted average card | Displayed in the top right when there are results, it shows the 전체 코호트 가중평균 리텐션, summed over all cohorts, by D-day. |
| Results table | Displays dimension columns along with cohort size and per-D-day retention (user count # and rate %) columns in table form. Rate cells are shown as a heatmap that darkens according to the value. |
4. Results Table Column Layout
| Column | Description |
|---|---|
| Dimension column | The dimension selected in grouping (channel, campaign, ad group, creative, platform, cohort date). It is pinned to the left and follows along when you scroll horizontally. |
| Cohort | The number of unique users grouped into that row's cohort. |
D{n} # | The number of users who returned (performed the return event) on that D-day. |
D{n} | The retention rate (%) for that D-day. The darker the color, the higher the retention rate. |
For each selected retention day (D1, D3, D7, D14, D30), a user count (#) column and a rate column are displayed together.