Metric Types

When you create a metric, the Field Aggregation dropdown in 2. Metric Configuration decides how the warehouse aggregates rows. This value is what the Metrics Catalog shows in its Metric Type column.

Two different fields are both called "type"

On the create form, the Metric Type radio (Non-MAB Experiment / MAB Experiment) selects the experiment family, not the math. The aggregation lives in Field Aggregation, and that is what this page covers.

Field aggregation options

Option	What it computes	Typical use
Average	The mean of a numeric value over exposed users.	Average revenue per user, average session length.
Sum	The total of a numeric value.	Total revenue, total ad impressions.
User Count	The count of distinct users meeting a condition.	Users who triggered event X.
User Proportion	The share of exposed users for whom an event occurred.	Conversion rate, click-through rate.
Ratio	A ratio between two aggregated quantities.	Revenue per session, events per user.

ABetterChoice also ships purpose-built aggregations for retention and activity:

Option	What it computes
Retention - All Users	Retention measured against all exposed users.
Retention - Unique Users	Retention measured against unique retained users.
Avg. active days per user	The average number of active days per exposed user.

Picking an aggregation

• Pick User Proportion when you care whether users did or did not do something.
• Pick Average when you care about the typical value a user generates.
• Pick Ratio only when neither the numerator nor the denominator alone tells the full story.
• Pick Sum carefully — outliers dominate sums and inflate variance.
• Use the Retention - * aggregations for retention curves and Avg. active days per user for engagement, rather than rebuilding them from User Proportion.