Feature vs Audience vs Dimension
These three objects all describe user attributes, but they answer different questions and live in different parts of the product. Use this page to pick the right one.
Quick comparison
| Concept | Lives under | Answers the question | Used at |
|---|---|---|---|
| Audience | Audience | "Who is eligible to enter this experiment / config?" | Targeting, before exposure |
| Feature | Data Management → Feature List | "Which user attribute may the model split treatment effects on?" | HTE analysis, after exposure |
| Dimension | Data Management → Dimension | "How do I manually slice the result?" | Explore analysis, after exposure |
How they differ
- An Audience decides who gets in. It is evaluated before assignment and gates exposure.
- A Feature (bundled into a Feature Combination) is a modelling input. It tells HTE (Heterogeneous Treatment Effect) Analysis which attributes — optionally binned via
Interval/Discrete/Manualgrouping — the estimator may use to find sub-populations with different effects. You do not pick the cuts yourself; the model searches them. - A Dimension is a manual cut for the Explore view. You choose the dimension and the platform reports the metric per value or per dimension group.
Practical guidance
- Reach for an Audience when you are choosing targeting.
- Reach for a Feature when you want the model to discover where an effect is stronger or weaker (HTE).
- Reach for a Dimension when you already know the cut you want to inspect (Explore).