gads_propose_create_experiment

Run a bid-strategy change as an A/B EXPERIMENT on a Search campaign instead of a full cutover (e.g. test Maximize conversions → Maximize conversion value without resetting the whole campaign's learning). Splits traffic (default 50/50) for ≥14 days; end or promote it later. Review-tier. The validate_only dry-run covers only the experiment shell — arm/schedule errors surface at apply time.

Type: money-moving write. Every write is dry-run validated, then classified by the safe-apply policy — auto-applied when bounded and low-risk, otherwise queued for approval.

Parameters