Interpreting General Performance Results
Often times when investigating a metric, the most important step is understanding what’s driving the performance and which subpopulations might be skewing this KPI. To accomplish this step, add a new analysis and click “Run”. Sisu will return key statistics for the metric, as well as min, max, median, average, sum, and the number of rows to help validate the quality of the data used. At any point, “Run” can be selected again to refresh these results, such as when new data becomes available.
The “Top drivers” tab will help you identify what is meaningfully impacting metric performance so that you can take action with confidence. Each line of the table represents a “Subgroup”, or a subset of rows in your data, that had a statistically significant impact on the metric.
- The name of the subgroup describes which column and value pair it refers to. For example, “gender = male” refers to rows where the value in the column “gender” is equal to “male”.
- The subgroup average describes the average for that specific subgroup, or subset of rows in the data. That column will reflect the calculation type that was selected, so it could be a Subgroup sum, average, or count. The label with the percentage change right next to it describes the relative percentage difference between this subgroup and the overall metric.
- The subgroup size describes how big that subset of the data is as a percentage of the total. For example, if there are 100 rows in the data and the subgroup size is 40%, it means that 40 rows out of the 100 match that criteria.
- The impact column describes how that subgroup impacted the overall metric performance. For example, if the overall “Average order value” is 100, a subgroup may have an impact of 20, meaning that having that subgroup in the data is increasing the average by 20. While another subgroup may have an impact of -10, meaning that having that subgroup in the data drives down your average order value by 10.
- Note: subgroups may overlap. This means that if all of the impact numbers from the different subgroups are added up, it will not get the overall “Average order value”.
There are multiple actions that can be taken for each fact in the table. These can be accessed by hovering over the “...” next to each fact. These include:
- "Hide fact" hides the fact for everyone. All hidden facts can be shown by clicking on the table settings and toggling “Show hidden facts”.
- "Add comment" enables the addition of commentary visible to other users
- " Save to fact board" enables users to save interesting facts to a fact board that is accessible on the bottom left side. Users will often use this feature to create a story for their stakeholders across multiple metrics that is easily digestible. Other users can then view the fact board and leave additional comments.
In addition, clicking on a given fact opens a window that can be used to dive deeper into a fact and check its relevance and importance.
Note: Also see Exploring a fact