Skip to main content

Optimizer

📄️Run Bid Optimizer

Runs the bid optimizer now for the given ad accounts, outside the nightly schedule. Runs asynchronously as a tracked background job; poll the returned `tracked_job_log_id`. Per account, the optimizer refreshes the bid-model estimates and proposes goal-based bid changes (plus ad-rate changes on eBay and target changes on TikTok and Google). What happens next depends on the effective optimizer mode: `off` skips the run, `monitor` stores the estimates only, `suggest` creates a pending change set that must be approved, and `auto` submits the change set for automatic application. Products in the control arm of a holdout experiment keep their bids. Accounts whose data is stale are skipped.

📄️Backtest Bid Model

Starts an offline backtest of the bid model over the given ad accounts: the model is trained on the first half of the history and scored on the second, comparing its calibration and contribution against the bid-to-target-POAS rule. Runs asynchronously as a tracked background job; poll the returned `tracked_job_log_id`. The result is stored and shown as `backtest` in Get Optimizer Settings; a passing result unlocks the optimizer's `auto` mode for the validity window (30 days by default).