Data-forward companies harness the power of their data by defining a set of “metrics” -- or key performance indicators (KPIs) -- and then analyzing them in different ways to make operational and strategic business decisions. 

Related Articles:

• Understanding Core Concepts in Sisu
• Understanding Impact
• Understanding Facts & How They Are Grouped

Understanding Metrics

A Metric is an aggregated calculation of a column within a data set (table or saved query from a data connection).

The Sisu workspace is centered around Metrics. Metrics can be defined at any time, and can be used by the entire company in creating Analyses.

Here’s a good way to think about Metrics: 

Metric Column + Aggregation Method = Metric

Using different Aggregation Methods (which are calculations such as average, sum, count, or rate) on a selected column in your data can produce different types of Analyses, for example:

• average order value (using an average calculation as the Aggregation Method on the column)
• total revenue (using a sum calculation as the Aggregation Method on the column)
• and more. 

In the following example, in the dataset the “order_value” column has been chosen as the Metric Column to analyze. This column can be used to define different Metrics, using various Aggregation Methods .

Note that in this example, the “AVG of order_value” is the average order value, which is the KPI that is being analyzed by Sisu with this defined Metric IF “Average” is selected as the calculation type Aggregation Method. If “Sum” is selected as the Aggregation Method for the Analysis, Sisu will perform the analysis based on the total order value (“SUM of order_value).

With different combinations of Metric Column and Aggregation Methods, you can perform a variety of General Performance Analyses. When you add “comparison” to an Analysis, you can perform useful Time Comparison and Group Comparison Analyses.

Metric Types:  Numerical and Categorical

A Metric is either numerical or categorical in nature. 

Numerical Metrics use Metric Columns that are numeric in nature (such as average order value, discount amount, number of days to close, etc.) and a numerical Aggregation Method for the Analysis.

The types of calculations that can be performed with numerical metrics are:  average, sum, and count.

Refer to the chart below for details about these calculation types:

Categorical Metrics use Metric Columns that are usually categorical in nature (such as any yes/no value, true/false values, SKU categories, etc.) and a categorical aggregation method (rate or count) for the Analysis.

Note that Metric Columns that contain a numerical value can also be used for Categorical Metrics. For example, you might want to analyze orders that contain one (1) item (e.g., Order Count = 1).

The types of calculations that can be performed with numerical metrics are:  categorical rate and row count..

Refer to the chart below for details about these calculation types:

What Sisu Does with Metrics

Sisu directly connects to data warehouses. An analysis can be built one of two ways:

1. On an existing table in the data warehouse
2. Through a custom query used to define Metrics using the Query Editor.
   (Any member of your organization can then investigate these metrics without needing to create or review the same SQL code every time.)
   
   

   Refer to the “Connecting & Managing Data” section for details about uploading or connecting to your data. Refer to Creating & Managing Queries for details about creating a custom query on your data.

With a metric column in a table defined, Sisu can run three types of analysis:

1. General Performance Analysis
2. Time Comparison Analysis
3. Group Comparison Analysis

Refer to Defining a Metric for instructions on selecting and defining a Metric to use in your Analyses. Refer to Understanding Sisu's Analysis Types for detailed explanations of each analysis type.