Sisu automatically generates Smart Waterfall Charts in Time Comparison and Group Comparison Analyses as a way for you to visualize your data. These charts are available in the Waterfall tab, next to Top Drivers and Query Preview.
- Interpreting a Time Comparison Analysis
- Interpreting a Group Comparison Analysis
- Understanding Impact
Understanding Waterfall Charts
Waterfall Charts visualize the drivers of change between two time periods or groups. Sisu automatically generates Waterfall charts by selecting a sequence of subgroups from the Top Drivers or All Subgroups page that best explain the change incrementally in your metric.
Sisu provides these Waterfall Charts with all Time Comparison and Group Comparison Analyses to help to distill, visualize and share up to 8 subgroups that best explain why your metrics are changing.
They also help you visualize and account for any overlapping impact between subgroups so you can see the incremental impact of each subgroup only.
|The waterfall chart is limited to visualizing the impact for each subgroup and does not help visualize or better understand change in the subgroup’s size or metric value, which is better analyzed using the Top Drivers table in the first subtab of your analysis. These values are still shown as text in the waterfall visualization for your awareness.|
The Sisu engine builds this chart by selecting the subgroups in your dataset that best explain the change in your metric. You can configure the chart to reflect your business logic. While the Fact table shows a list of key subgroups driving your metric, with waterfall charts you can visualize the impact of a few selected subgroups and make it easier for your team to discuss insights and make decisions.
The following image is an example of a Waterfall Chart for a Time Comparison Analysis that analyzed the Average Order Value change between two time periods.
The chart includes the two time periods’ average order value on the two ends of the chart, with the previous time period (or Group A in a group comparison analysis) at the far left, and the recent time period (or Group B) at the far right.
In between these two bars, additional bars are displayed representing facts from the analysis that have incremental impact on the metric’s change in value between the two time periods or groups. Facts with positive incremental impact are displayed in green, and facts with negative incremental impact are displayed in red. A toggle at the top allows you to view overlapping impact.
For example, in the chart, we can see that orders in Los Angeles impacted the change in sales negatively by 0.82, while orders in New York positively impacted sales by a value of 0.43. Understanding these various facts in between the change can point to valuable insights for your operations and business.
|For a greater understanding of “impact” and what the impact values mean, refer to Understanding Impact.|
Hover over any bar in the chart to view that fact’s details, including Total impact, incremental impact, overlapping impact, change in average, and change in size:
Each of these areas is described below:
The overall or net impact of the subgroup across all the rows defined by the subgroups. It is independent of any other subgroup in the Waterfall chart and is the same value for impact shown in the Top Drivers or All Subgroups tab for this same subgroup.
Total Impact is always the sum of the Incremental and Overlapping impact.
The solid (red or green) part of the bar that shows the impact of this subgroup after accounting for any of its rows whose impact was already captured in the subgroups before it in the sequence of subgroups.
Incremental impact is the impact that is left over after removing any overlapping impact from the subgroup’s total impact. The number shown at the top of each bar in the Waterfall chart is this Incremental impact of each subgroup.
The shaded part of the bar that shows the impact of the rows in this subgroup that were already captured in the subgroups before. Refer to About Overlapping Impact section for more details.
You can toggle the view of overlapping impact on or off based on your needs.
Describes the change in metric value for the Subgroup. Can be Average, Rate, Sum, Count, depending on the Aggregation Method used for the Analysis.
Change in Size
Describes the change in size of this population between the two comparison points.
For example, “this population was 23.7% of the total rows in Period 1, and is now 26.8% of the total rows in Period 2.”
About Overlapping Impact
Each subgroup is comprised of a number of rows in your data-table. Each row of data can have a positive or negative impact to your overall metric. The total (or net) impact of a subgroup is the sum of all these individual positive or negative impacts across each row in the subgroup.
Positive impact is displayed in green, and negative impact is displayed in red. When the overlapping impact is the same polarity (for example, green over green), that means that the subgroups before it in the Waterfall chart captured rows that collectively had the same positive impact, and that positive impact is compounded.
In some cases, the bar will show overlap with an opposite polarity. When a subgroup has a positive incremental impact but negative overlapping impact, it means that the subgroups before it in the Waterfall chart captured the rows that collectively have a negative impact (i.e. negative overlapping impact) and that the remaining rows had a positive incremental impact.
In the example shown above, the indicated subgroup has a negative incremental impact but a positive overlapping impact (the solid bar is red and the shaded portion is green). The previous subgroup in the waterfall chart captured the positive impact of this subgroup and, after accounting for them, this subgroup has an incremental negative impact.
The inverse is also true: when a subgroup has a positive incremental impact but a negative overlapping impact (the solid bar is green and the shaded portion is red), the previous subgroups in the waterfall chart captured the negative impact of this subgroup and, after accounting for them, this subgroup has an incremental positive impact.
|Refer to Understanding Impact for a full explanation of Impact.|
Customizing a Waterfall Chart
Sisu automatically generates the chart that shows the key subgroups with the largest incremental impact that drove your metric. However, you may want to customize the chart’s settings and subgroups to reflect your business logic or to answer a specific question.
For example, for an ecommerce company, you may be first interested in seeing how different states/regions drove the net change in your business before evaluating further factors.
You can add or delete subgroups to the chart, and reorder the subgroups.
The chart will update automatically with new calculations based on your customizations when you rerun the analysis.
Any setting changes you have made will be persisted for the next refresh. If a certain setting cannot be persisted (for example, the column you requested is no longer part of the data set) then the chart will be generated based on all the other settings that can be persisted.
Adding a Subgroup
To add a new Subgroup to a Waterfall Chart:
- Click Add a Subgroup button.
This will be enabled as long as you have less than 8 subgroups in your chart already.
- Choose to add subgroups from the Top Drivers table or define your own custom subgroup by combining factors from your dataset.
- Click Add new subgroup.
- Click Refresh chart
Deleting a Subgroup
To delete a Subgroup from a Waterfall Chart:
Hover over the subgroup name and click on the delete icon that pops up.
To move a Subgroup within a Waterfall Chart:
Hover over the subgroup name and use the “6 dots” icon to drag-n-drop the subgroup to the location you want in the chart.
Resetting a Waterfall Chart after Making Edits
After you RUN the analysis again the chart will be reset to its default sequence using the appropriate Sisu algorithm.
Subgroups with “Zero” Incremental Impact
This means that the entire impact of this subgroup was already captured in subgroup’s to the left of it. For example if your first subgroup was Country=USA and the second one was State=California, this second subgroup is likely to have a zero incremental impact since the first subgroup already captured it’s impact since all the rows for State=California are also Country=USA.