​
What Are Numeric Metrics?

Numeric metrics aggregate event values per user—counting occurrences, summing totals, or averaging amounts—to measure feature impact.

Unlike conversion rates, they focus on raw values (e.g., purchase prices, session times).

Equal weighting across users in Experiments is ensured by averaging data per user, preventing heavy user bias from skewing results.

​
Aggregation Methods

Numeric metrics require an aggregation method to process event data:

Count

The average count of events per user.

Example: Average Purchase Count per User

  • If User A makes 3 purchases and User B makes 1, the metric averages to 2.
  • (3 + 1) / 2 = 2

Use Case: See if a feature increases purchase frequency (e.g., via a new layout).

Returning users raise the average, new users lower it.

Other examples: The average number of video plays, help requests per user.

Sum

The average sum of event values per user.

Example: Average Total Spend per User

  • If User A spends $100 and User B spends $50, the metric averages to $75.
  • ($100 + $50) / 2 = $75

Use Case: Test if a feature boosts total spend (e.g., via upselling).

Frequent users increase the sum, new users dilute it.

Other examples: The average total time spent watching videos, pages visited on your site.

Average

The average of per-user event averages.

Example: Average Purchase Price per User

  • If User A’s purchases average $40 and User B’s average $60, the metric is $50.
  • ($40 + $60) / 2 = $50

Use Case: Check if a feature changes typical purchase value (e.g., via a discount).

The average reflects per-user values—new users’ data can shift it up or down based on their behavior.

Other examples: The average time on site per session, items added to the cart.

​
Next Steps