Smart Tuning (Calibration)
Smart Tuning calibrates threshold behavior using measured outcomes and exposes calibration status, thresholds, and recent runs.
Updated 2026-02-21
Smart Tuning calibrates threshold behavior using measured outcomes and exposes calibration status, thresholds, and recent runs.
Prerequisites
- Connected Amazon account
- Selected marketplace
- Minimum of 2 calibration reports before activation
Expected outcome: You can evaluate and activate calibration only when readiness criteria are satisfied.
Quick Start
- Open
Settings > Smart Tuning. - Check calibration status and report count.
- If report count is at least 2, activate Smart Tuning.
- Review calibrated thresholds table.
- Track recent calibration runs.
Expected outcome: Calibration is enabled with observable evidence and history.
Detailed Workflows
Workflow: Readiness Check Prior to Activation
- Confirm report count and calibrated type count.
- Verify latest report timestamp.
- Ensure account context is correct.
- Activate only when readiness is met.
Expected outcome: Activation occurs at the right maturity stage.
Workflow: Activation and Validation
- Click Activate Smart Tuning.
- Confirm status changes to active.
- Verify thresholds table populates with calibrated values.
- Compare baseline vs calibrated values by type.
Expected outcome: Calibration state and threshold outputs are validated immediately.
Workflow: Ongoing Calibration Review
- Check recent run list each review cycle.
- Monitor total samples and calibrated type coverage.
- Track average improvement indicators.
- Adjust broader automation policy if calibration quality changes.
Expected outcome: Calibration performance is monitored as part of routine governance.
Workflow: Deactivation for Controlled Rollback
- Deactivate Smart Tuning when needed.
- Confirm status changes to inactive.
- Keep historical reports for future reactivation analysis.
Expected outcome: Calibration can be rolled back without losing historical context.
Common Errors
Error: Activation blocked by readiness requirement
- Verify report count.
- Continue accumulating outcomes.
- Retry activation after next calibration cycle.
Expected outcome: Activation proceeds once minimum readiness is reached.
Error: No thresholds visible after activation
- Verify calibrated types count.
- Check recent report output.
- Wait for next run if data is insufficient.
Expected outcome: Missing thresholds are traced to sample coverage, not random UI behavior.
Error: Calibration appears active but no impact observed
- Compare baseline vs calibrated threshold values.
- Validate recommendation types in current queue match calibrated coverage.
Expected outcome: Operator can confirm whether calibration scope matches active workload.
FAQ
Is Smart Tuning mandatory for Auto-Apply?
No. It is optional and intended for data-mature accounts.
What hard requirement exists for activation?
At least 2 calibration reports.
Can I deactivate and reactivate later?
Yes, calibration state can be toggled without deleting historical reports.
Last Updated
- Last updated: 2026-02-21
- Version assumptions:
CalibrationSettingsPageand/settings/calibrationstatus/reports/thresholds handlers
