seasonal-promo-swap

Use menu engineering quadrant analysis to find star items (high popularity, high margin) and cash cows (high margin, low popularity) as promotion candidates. Exclude dead items and out-of-stock products, then rank remaining candidates by actual margin rate rather than intuition.

Analyze the hourly heatmap to identify peak time slots, then assign lighter lower-priced items to lunch peaks and high-margin items to dinner peaks. Create visibility schedules for each time slot after owner approval, one operation at a time.

No, the price change simulation only checks price tolerance and estimates expected margin increase from the promotion. It is not a demand forecast and should not be presented as a prediction of future revenue.

If inventory filtering removes all candidates, the Skill honestly recommends keeping the current lineup for the week rather than forcing a swap. It never fabricates candidates to fill an empty list.

No, all execution operations such as creating promos, reordering menu items, and scheduling visibility require explicit owner approval first. Each approved action runs as a single operation, never bundled together.

Campaign ROI is measured several days to a week after execution using lift and holdout comparison. If the lift is not positive versus the holdout group, the Skill rolls to the next backup candidate instead of leaving an ineffective promotion in place.