You are here:
Primary Metric Results
The third section of the dashboard shows results specific to the primary metric selected for the experiment. Results for the primary metric determine the experiment's winner.
Metric analysis focuses on content and conversion rate, providing totals, lift (difference), and chance-to-win metrics.
| Column | Description |
|---|---|
| Cohorts | The names of each experiment cohort. |
| Personalizer | The name of the recommender tested for this cohort. If a recommender isn’t used for a cohort, the Personalizer table cell is left blank. |
| User-Defined Primary Metric | This column name varies depending on the primary metric defined for the experiment. For example, this column can report revenue or recommendation click-through rate. |
| Lift | Percent difference of the tested metric when compared to the control or across all variation cohorts. |
| Credible Interval | A credible interval is the probable range in which the true value of a metric falls, based on the observed data. |
| Chance to beat control | Probability that a cohort’s content is better than the control, based on the observed data. Multiple variation cohorts can beat the control. |
| Chance to beat all | Probability that a cohort’s content is better than every other content in the experiment, based on the observed data. Only one variation cohort can beat all. |
Did this article solve your issue?
Let us know so we can improve!
