CASE is a conditional statement that performs logical tests on a series of conditions and returns values when a condition is met. The CASE function returns a numeric result. You use CASE for complex conditional expressions that contain two or more conditions.

CASE is followed by WHEN/THEN keywords to specify the conditions, then by ELSE to specify the default return.

If no match is found, the default return expression (ELSE) is used. If there is no default return and no values match, then null value is returned. All CASE expressions conclude with the END keyword.

CASE and Multidimensionality

Before proceeding to the CASE syntax samples, as a user you have to understand that usage guidelines for CASE in the MAQL environment are based on multidimensionality. Understanding multidimensionality is crucial to ensure your CASE metrics return correct results (see MAQL and Multidimensionality ).

  • CASE does not work with the base data level.
  • Instead, the condition is evaluated on the aggregation dimension.
  • All WHENs are evaluated independently but all based on a single dimension (the aggregation dimension mentioned in the previous point).

In most common scenarios you should follow this rule:

  • Always put the CASE statement inside (not outside) an aggregation (SUM) container.

See examples below.

CASE Syntax

CASE can be used in the following two syntax forms:

No aggregationSELECT CASE WHEN condition1 THEN outcome1, WHEN condition2 THEN outcome 2 ELSE outcome3 ENDSELECT CASE WHEN {attribute/activity_type} IN ("email", "meeting") THEN 1, WHEN {attribute/activity_type} IN ("phone_call") THEN 2 ELSE 0 END
With aggregationSELECT SUM (SELECT CASE WHEN condition1 THEN outcome1, WHEN condition2 THEN outcome2 ELSE outcome3 END)SELECT SUM (SELECT CASE WHEN {attribute/activity_type} IN ("email", "meeting") THEN {fact/amount} ELSE {fact/loss} END)

Using the aggregation container ensures that the CASE conditions are computed on the correct dimension and returns relevant results.

How to Use CASE Correctly

As noted above, we advise you to use CASE inside aggregation containers.

Here is an example of incorrect use of CASE in the following logical data model. For the sake brevity, only one CASE condition is used.

An example of incorrect use of CASE

If you want to slice the following metric with Campaign Category, the insight would not compute because the WHEN conditions are not related to insight dimensionality:

SELECT CASE WHEN {fact/customer_state} IN ("AR", "NY") THEN SUM ({fact/price}) ELSE SUM({fact/price} * {fact/quantity}) END

On the other hand, the same metric when sliced by attribute Customer ID (instead of Campaign Category) is valid. That is because the slicing attribute (Customer ID) is directly related to the CASE condition attribute (Customer State).

If you need to slice by unrelated attribute, put the CASE statement inside the SUM aggregation instead:

SELECT SUM(SELECT CASE WHEN {attribute/customer_state} IN ("AR", "NY") THEN {fact/price} ELSE {fact/price} * {fact/quantity} END)