Drill down hierarchy

Drill down hierarchy is in beta 🧪 right now and unactivated by default. Please send us any feedback to get this improved!


General usage

The drill down input can be defined in Settings > Drill down hierarchy. It allows to define the methodology to break down main KPI variation. 3 options are available:

Once activated (i.e. not defined to “Steps then Dimensions”), DataMa uses what is calculated as the most interesting drill down. Yet, you can manually change the drill down by right clicking on a total bar of the waterfall and selecting the desired drill down

Available drill down hierarchy

Steps then Dimension (default setting)

By default, the main entry point for Compare analysis are the steps of your market equation, a.k.a. metric relation. Then, within each step, DataMa helps you find what is the most interesting dimension to explain the gap of a given steps. This is the default way of working of DataMa Compare, as specified in (Model details)[https://datama-solutions.github.io/docs/docs/core_app/new/compare/model/model.html]

Dimension then steps

When your main KPI is the result of a simple computation (i.e. product or sum of metrics steps), DataMa is able to breakdown the impact of each segment of each dimensionon the main KPI.

This is particularly interesting in cases like

By activating the “Dimensions then Steps” option, you will see the impact of the top 5 segments of the most interesting dimension for the total step variation to explain the gap. Then, by clicking on a given segment, you will be able to get the details of the impact of each step of your market equation on that specific segment.

The title and comment of the first slide of DataMa Compare are also updated to use the most interesting dimension wording

Few tips to consider

Auto

Auto mode lets DataMa score the interest of splitting by “Steps then Dimension” vs. all the different variants of “Dimensions then Steps” using an interest score. You can see the interest score by right clicking on the total bar and reading the contextual menu. While the absolute value of the interest score has no much interest, the ranking helps in finding the proper approach for explaining the variation. DataMa considers the share of the total gap explained by one specific step vs. the share of the total gap and relative variation of the biggest segment to create that score.

Few notes