All Analyses run in Sisu require a key metric on which to base the Analysis.
- Understanding Metrics
- Editing a Metric
- Deleting or Removing Metrics
- Editing a Metric's Underlying Query
Defining a New Metric
A Metric can be defined by either using an existing table in the data warehouse or by writing a custom SQL query in the Sisu Query Builder. Once a Metric has been defined, it can be used within multiple Sisu projects.
To create a new Metric:
- From the Projects module, select the desired Project and then click + (plus sign) in the Metrics panel on the left.
The Add metric modal will be displayed.
Note that any existing Metrics that you or your colleagues have previously created will appear in the Metrics list. These are Metrics that have been defined but have not been added to the current project.
If the Metric that you need is in this list, you can select it and click Add metric instead of creating a new one.
You can also edit or delete a Metric and edit the Metric’s underlying query from this modal.
- Click Create new, and complete the requested fields as described below.
Type a name for your new Metric. We recommend being specific and accurate so that any teammate searching for that metric will easily be able to find it. Some users prefer to add a version to the Metric name (e.g., Average Order Value v1).
Select where the data lives. CSVs can be used, but we recommend a live connection to a data warehouse to ensure every team member is always working with fresh data.
Table or saved query
Select the name of the table or the name of the custom query previously created for usage with this metric. For more information on this, please refer to Editing a Metric's Underlying Query.
Select the column that represents the KPI (e.g., a revenue column or churn flag). The metric can be numerical or non-numerical (categorical).
If a non-numeric metric (such as loyalty status or churn status) is selected, a Select category field will be added.
This field is only visible if a non-numeric, categorical Metric was selected in the Metric column field. Select the appropriate value to define the category (e.g., “yes” or “no”)
This field is optional but we recommend using it to help perform more advanced analysis, like a time comparison between two periods.
Click to add filters, if desired.
For example, if your data set includes different order channels, you may be interested in analyzing on digital mobile orders separately.
In that case, add a filter, select the column (in example, “ORDER_CHANNEL”) equals value “digital_mobile”.
Or maybe you may want to look at user churn in the first 30 days. In that case, you may select a column like “days_since_signup”, select “<=” as the operator, and select “30” from the value column.
The filter function supports multiple operators: =, <=, >=, <, >, !=. You can also add additional conditions with the AND / OR functions as well as filter by certain keywords within your data. Refer to Setting Nested Filters and Transforming a Column: Keyword Analysis for more information.
Select “number” if your Metric is numerical (e.g., revenue, or customer age).
Select “category” if your Metric is categorical (e.g., order type or churn status), and then ensure a category is selected in the Select category field.
Specify what type of calculations should be done with this metric. You’ll be able to edit this for every analysis, but this will set the default value.
Note that the calculation options are different for numeric and categorical Metric types.
- Select “Average” for an average order value metric.
- Select “Sum” for total sales related metrics.
- Select “Rate” for a categorical metric such as churn rate, to analyze the percentage of users that have churned.
- Select “Count” for a categorical metric such as churn rate, to analyze the total number of users that have churned.
Specify whether you want this metric to increase or decrease.
This is an important step because if your goal is to decrease a metric, every time we find a driver that increases that metric, Sisu will display it as a negative impact in red. Conversely, if your goal is to increase a metric, drivers that increase that metric will be displayed in green to indicate a positive impact.
Here is an example of a completed screen:
- Click Create and the Metric will automatically be added to the project.
A Metric can only be added once to any project, but it can be added to multiple projects.