restaurant menu engineer

The problem

Restaurant owners often rely on gut feeling to decide which dishes to keep or cut, leading to "Plowhorses" that drain margins and "Puzzles" that remain under-ordered despite high profitability.

What it does

• Classifies every menu item into the Stars, Plowhorses, Puzzles, and Dogs matrix using actual POS sales-mix and recipe cost data.
• Calculates precise fair-share popularity and weighted-average profitability thresholds instead of using subjective impressions.
• Identifies specific items requiring price increases, cost reductions, or promotional pushes based on their quadrant.
• Rewrites menu descriptions using cited behavioral research, such as Cornell's findings on descriptive naming and dollar-sign removal.
• Flags data gaps where units-sold or plate-cost figures are missing to prevent inaccurate classifications.

Why this beats prompting it yourself

General LLM prompts often hallucinate sales figures or use generic "copywriter" instincts that ignore actual restaurant economics. This skill enforces strict data integrity, requiring real POS exports and citing specific academic research for every description change rather than making unverified marketing claims.

Use cases

• Conducting a quarterly menu audit to identify low-margin items that need a recipe rework.
• Optimizing "Puzzle" items by applying research-backed sensory descriptions to increase order frequency.
• Determining which seasonal items to keep permanent based on their performance against the fair-share threshold.
• Pre-launch vetting of new menu descriptions to ensure they avoid illegal sourcing claims while maximizing appeal.

Known limitations

Requires actual units-sold and plate-cost data to function. It does not provide graphic design, layout services, or final price setting.