You are here:
Formulas, Operators, and Calculated Insights
Create audiences in different ways and at different stages of your workflow using formulas, operators, and calculated insights in Data 360. During ingestion, you can use formulas to perform operations on row-based data used downstream. For segmentation, audiences are created with segment operators.
With calculated insights, you can use views to understand large-scale behavioral data and reuse views to enhance segmentation. You can also find insights in business intelligence tools. Which option you use to perform a data transformation depends on your use case and the type of data.
Formulas
A formula is good for calculating different data types but requires advanced knowledge. A formula is derived from other source fields, but you can use hardcoded values.
It’s recommended to use a formula when a simple row-based operation is used to abstract logic that the user can consume and use at segmentation time. A row-based calculation applies logic to self-contained rows as opposed to aggregating across rows. For example, an IF formula that changes “Shoe” to “Shoes” works because the result is static. But finding the number of days since January 1 doesn’t work because the number of days changes over time. The calculation occurs only at ingestion time and is a stored calculation, not a run-time calculation. Formulas don’t automatically update.
This table summarizes when to use a formula.
| Use Case | Formula |
|---|---|
| Simple logic on a row-based operation |
|
| Ease of use and self-service |
|
| Highly reusable content |
|
| Attribute updated regularly |
|
Segmentation Operators
Use an operator to complete self-service filtering use cases without the help of a data analyst. Segmentation operators are a good choice for:
- Simple aggregations, like count, on one object
- Max, min, avg, and sum aggregations on numbers when the conditions are simple and unlikely to be reused often
- Filtering on 2 years or less of aggregation data
If segmentation operators don’t fully support your use case, use a calculated insight instead. For example, the max operator on dates is available only with a calculated insight, for example, finding the last purchase date of shoes. Use a calculated insight for max, min, avg, and sum aggregations on numbers that are reused among segments or if they have a long list of other attributes as part of the filter container.
This table summarizes when to use a segmentation operator.
| Use Case | Segmentation Operator |
|---|---|
| Ease of use and self-service |
|
| Highly reusable content |
|
| Attribute updated regularly |
|
Calculated Insights Functions
To understand customer behavior, use a calculated insight at the customer or entire population level with multiple dimensions, such as customer, channel, and product attributes. You define and calculate multidimensional metrics, such as recency, frequency, monetary value (RFM), and customer lifetime value (CLV). After the metrics are calculated, they can be automatically refreshed daily so that you’re operating on the most recent information for your channel, customers, or products.
Calculated insights make it easy for users to define segment criteria and personalization attributes for activation using metrics, dimensions, and filters. Calculated insights are best used for:
- Nontrivial calculations, such as calculating a net promoter score as a percentage
- Complex queries across multiple objects
- Reusability, such as when you expect a view to be used multiple times across different values or when combining related attributes in the same view
For example, you want to find web page views and page clicks that belong to the same view and a product category that changes as a dimension. If implemented using segmentation, you must redefine the conditions every time there’s a change. With a calculated insight, after the dimensions are set on the view, you can reuse the dimensions as needed. Because you can only have a maximum of 50 measures and 10 dimensions, you can organize the view to have different views and group metrics together on the relevant object. If possible, avoid having a view include different channels in which all the engagement is in one cross-channel view.
This table summarizes when to use a calculated insight.
| Use Case | Calculated Insight |
|---|---|
| Simple logic on a row-based operation |
|
| Ease of use, self-service |
|
| Highly reusable content |
|
| Attribute updated regularly |
|
