Narrative Settings: Drivers
Sales increased by $23.8k month-over-month, but why? What drove that change? Did anything actually offset that increase with a negative impact to sales?
Driver analysis, sometimes referred to as contribution analysis or attribution analysis, helps explain the why behind the facts and figures that are called out in the narrative. You'll notice a handful of insights in your discrete and continuous narratives that point out this kind of information, making it easy for the reader to understand exactly what's going on in the data and why.
Discrete example explaining total sales - The total sales of $2.1 million was driven by Technology with $764,250 (36%), Furniture with $673,536 (32%) and Office Supplies with $657,138 (31%).
Continuous example explaining change in total sales over the last month - The $7,451 increase in sales over the most recent month was driven by Furniture (+7.34% from $84,229 to $90,411) and Office Supplies (+5.31% from $68,568 to $72,206) and was offset by Technology (-2.92% from $81,216 to $78,848).
Driver analysis is configurable, allowing the user to select the method to be used (e.g. count, percentage, etc.), the threshold for how many drivers to call out, and also whether or not to include offsetters.
Driver Analysis Methods
- Count - specify the number of entities to be called out using the thresholds provided. For example, choose this method if you always want to see the top three.
- Individual % - specify a minimum threshold and the narrative will include any individual entity greater than that percentage threshold. The percentage threshold is relative to the total value being explained for each insight. For example, choose this method if anything less than 5% of the total is deemed immaterial.
- Cumulative % - specify the percentage of the total value that you wish to be explained. Entities will be listed in order of magnitude until the cumulative value of those entities explains x% of the total value. For example, choose this method if you want to know which entities factored into at least 90% of that total value.
Other Driver Settings
Offsetters are entities that move in the opposite direction and detract from the total value.
e.g. - The total profit of $117,107 was driven by Binders with $33,885, Paper with $30,803 and Storage with $18,500 and offset by Supplies with -$1,232.
Secondary Contributors are available when you have a second dimension (must be a non-time dimension). For each driver that is called out, it will go one level deeper and call out the drivers of that entity as well.
e.g. - The profit of $268,224 was driven by Technology with $134,218 (led by Copiers with $47,650 and Phones with $40,272) and Office Supplies with $117,107 (led by Binders with $33,885 and Paper with $30,803).
Use with caution - if thresholds are high, secondary contributors can be a lot of information to fit into a single sentence or bullet.
Verbosity is a setting in the Language tab, but adjusting this changes the way the driver content is written. You'll see more information in parentheses with higher verbosity, and you get a more concise version of the driver analysis with lower verbosity
For measures that are actually made up of other component measures or subcategory measures, driver analysis can explain what impact each measure had on the parent. Take Total Expenses as an example. It's made up of Operating Expenses and Non-operating Expenses.
The $2.1 million in total expenses was driven by operating expenses with $1.8 million and non-operating expenses with $335,188.
To configure metric drivers, you'll need to specify the relationships between each measure. In this example, I'm setting up two relationships:
- Operating Expenses as a subcategory of Total Expenses
- Non-operating Expenses as a subcategory of Total Expenses
Note - You must have multiple measures for metric analysis.