Sort Your Menu Into Stars, Plowhorses, Puzzles and Dogs From a POS Export With AI

Context: You are a restaurant, café, salon or shop owner reviewing which items earn money. Run it quarterly, or before reprinting a menu or price list.

Act as a hospitality and retail profit analyst. I uploaded our POS item-sales export for [DATE_RANGE] from [POS_SYSTEM].
Unit costs: [COSTS_SOURCE — e.g., "column 'Unit Cost' in the file" OR paste "Item: cost" lines here].

1. For each item, calculate units sold, revenue, unit cost, contribution margin per unit (price minus unit cost) and total contribution. Show the formula you used.
2. Classify each item within its category using menu engineering:
   - STARS: high popularity, high contribution margin
   - PLOWHORSES: high popularity, low contribution margin
   - PUZZLES: low popularity, high contribution margin
   - DOGS: low popularity, low contribution margin
   Use the category's average units sold and average contribution margin as the cut-offs, and state the cut-offs.
3. Flag any item with a missing or suspicious cost, and exclude it from the classification.
4. Suggest up to 3 Plowhorse tests (price change, portion or recipe tweak, or pairing) and say what to measure for [TEST_WEEKS] weeks. Don't predict the result.
5. Suggest 3 bundles pairing a Star or Plowhorse with a Puzzle, each with a one-line menu description in our tone: [BRAND_TONE].
6. List Dogs to review, noting any that might matter for other reasons (dietary options, kids' items).