Create a Menu Performance and Experiment Workspace in Enter for a U.S.-based multi-location restaurant group. Help menu, finance, and operations teams compare items on a like-for-like scope, explain why contribution changed, turn a proposed menu change into a controlled experiment, and launch it only after finance and operations approve.

[TARGET USERS]
- Menu Director: Review item performance, open item detail, and propose price, recipe, bundle, photo, regional, or removal experiments.
- Finance Analyst: Review ingredient cost, fees, discounts, and contribution effects, then approve or return an experiment.
- Operations Manager: Check ingredient availability, prep changes, training, POS updates, and store communication before launch.
- Administrator: Manage source data, metric definitions, stores, channels, and user access.

[CORE WORKFLOW]
1. Filter by current period, an equal-length baseline, region, store, category, channel, daypart, and promotion status. Echo the active scope and baseline on every page.
2. Compare units, net sales, ingredient cost, contribution after ingredient cost, contribution after ingredient cost and fees, food-cost percentage, and stockouts on the same scope.
3. Open an item to see the period comparison, channel breakdown, recipe cost changes, and related experiments. An AI analyst may draft questions and hypotheses from the visible metrics, but must cite evidence and never change prices or approve anything.
4. Create an experiment with item, action, stores, channels, dates, owner, hypothesis, success metric, and stop threshold. Record whether a concurrent control group and a significance test exist.
5. Keep Finance review and Operations readiness as separate approvals with checklists. After an approved launch, record 7-, 14-, and 28-day observations and decide to scale, iterate, roll back, or archive.

[CORE PAGES]
- Menu Dashboard: Scope filters, KPI cards with baseline changes, largest contribution increases and decreases, category performance, margin pressure, and data coverage.
- Item Explorer: Item list, item detail, period comparison, channel breakdown, supporting evidence, and linked experiments.
- Experiments: Experiment list and detail, interpretation limits, review status, readiness checklist, and observation windows.
- Approvals & Execution: Finance and Operations queues with recorded decisions.
- Data Management: Imports, metric definitions, mappings, quality warnings, and permissions.

[DATA AND BUSINESS RULES]
- Use stable IDs to link items, stores, channels, periods, sales facts, recipe costs, experiments, reviews, and observations.
- Net sales = gross sales - discounts - refunds. Contribution after ingredient cost = net sales - ingredient cost. Contribution after ingredient cost and fees = that amount - marketplace fees. Food cost % = ingredient cost / net sales. Neither contribution measure is net profit; handle zero denominators explicitly.
- Compare only like-for-like scopes; if coverage or definitions differ, label the comparison as not directly comparable.
- Label before-and-after differences as observations, not causal effects, unless a control group and significance test are recorded.
- A proposed change stays "not applied" until both approvals are recorded; never update prices automatically.
- Include a demo case: Harborline Kitchen, 24 stores, Aug 4-31, 2026 compared with Jul 7-Aug 3, 2026. Truffle Mac & Cheese sells 21,376 units in both periods with flat net sales of $362,323, but ingredient cost per unit rises from $5.25 to $7.15, so contribution after ingredient cost and fees falls by $40,614 (-17.3%). Experiment EX-F03 proposes a +$1.00 dine-in price test in 6 stores from Sep 8 to Oct 5, 2026, with no control group; Finance and Operations are both pending, so no result is recorded.

[VISUAL DIRECTION AND DELIVERABLES]
- Use a clean analytics workspace with compact KPI cards, readable tables, visible scope labels, and clear status badges.
- Support desktop and mobile, keeping filters and wide tables usable on small screens.
- Use labelled demo data with consistent IDs and dates, USD amounts, and U.S. date conventions.
- Deliver a publishable demo with one traceable path from item diagnosis to experiment approval.

Before building, provide a requirements summary, missing information, page architecture, data model, metric formulas, role permissions, state transitions, and acceptance criteria. Do not begin generating pages yet.