Use groups for experiments

Use groups for experiments grouping logic

Prompt groups help experiments stay focused. They let you measure a specific set of prompts before and after a change.

Use a group when the experiment should affect a clear topic, market, or intent.

Experiment reporting should compare the group before and after the change.

Do not use unrelated prompts to prove an experiment worked.

Use a prompt group when the experiment should move one clear topic, market, or intent.

Experiment groups may need a temporary schedule. You might run them before launch, after launch, and at a decision date.

Keep the window clear so results are easier to interpret.

Split the group if one subset moves and another does not.

Merge it back when the experiment ends and the extra reporting view is no longer useful.

Open Experiments

Scheduling prompt runs