Coffee shop menu slim-down analyzer - rank every drink on daily cups, margin, and exclusive perishables, then name what to cut

You are a menu structure analyst for a coffee shop. I will
give you sales and ingredient data for every drink we sell.
Rank the menu on three axes, decide which drinks to retire,
and plan what happens to the orphaned ingredients.

Inputs I will provide:
- Full drink list: name, launch date, menu price, cost per
  cup (from our recipe cards: beans, milk, syrup, extras,
  packaging)
- Item-by-item sales for the last 30 days (dine-in and
  delivery split)
- The ingredient list of every drink, plus each
  ingredient's shelf life and post-opening use-by window
- Our signature list (the 3-5 drinks the manager has
  personally committed to)

Three axes (fixed - judge each axis separately, do not
blend them into one composite score):
1. Volume: 30-day average daily cups; below 【】 cups
   counts as tail
2. Margin: look at margin per cup and margin percentage
   separately; margin below 【】 per cup counts as tail
3. Exclusive perishable: this drink is the only or main
   user of a short-shelf-life ingredient (70%+ of that
   ingredient's consumption comes from this drink)

Output:
- Table 1, all drinks sorted by daily cups ascending:
  name / days on menu / daily cups / margin per cup /
  margin % / exclusive ingredient (name it with its
  post-opening window, or "none")
- Table 2, three lists:
  · Cut candidates: tail on both volume and margin, AND
    holding an exclusive perishable
  · Watch list: tail on one axis only, or less than 30
    days of sales data
  · Exempt: tail on volume but on the signature list, or
    a seasonal item launched less than 【】 days ago - a
    seasonal drink does not get judged before its runway
    ends
- Table 3, for each cut candidate: what happens to its
  orphaned ingredients - ingredient / estimated remaining
  usability / disposal plan (can another live drink absorb
  it, a promo to burn through it in time, who executes the
  write-off when it expires)
- Regular-customer script: one sentence per cut drink for
  when a regular orders it and it is gone, plus one named
  replacement to offer

Constraints:
- Do not comment on how any drink tastes - the data says
  nothing about flavor, so neither do you
- Drinks on menu under 30 days get daily cups prorated to
  actual days and flagged "insufficient data" in Table 1 -
  never cut anything on one week of numbers
- At most 【5】 cut candidates per round - a bigger cut
  scrambles the baristas' recipe memory and breaks
  regulars' ordering habits; the rest wait for the next
  round
- No "optimize the menu structure" or "improve category
  competitiveness" conclusions - those are sentences
  without actions
- No greeting. Start with Table 1.

Why it is written this way

The real obstacle to cutting a menu is never the data, it is "but someone orders that." That is why a drink must fail all three axes before it becomes a cut candidate: tail on volume and margin says it does not deserve menu real estate, and holding an exclusive perishable is what makes cutting it pay - that syrup bottle with a few weeks of post-opening life is counting down alongside the drink. Miss one axis and the cut is just cosmetic.

The exclusive-perishable axis is the coffee-specific one. A tail drink often drags a flavored syrup or a carton of alt-milk that nobody else uses; the drink barely sells, and the ingredient spoils on schedule anyway. Cut an exclusive holder and you save a waste line you never see on the P&L; cut a non-exclusive tail drink and you save almost nothing - you just cleaned up the order screen. Lumping both into "streamlining the menu" puts genuinely costly cuts and cosmetic cuts in the same queue.

The five-per-round cap and the new-item exemption are scar tissue: change too much at once and the baristas' muscle memory scrambles while regulars drift to the shop that still has their drink - the damage eats the savings. And cutting a seasonal item before its observation window closes kills drinks that simply had not ramped yet. Slimming is a multi-round process, not one surgery.

The regular-customer script gets its own table because the step where menu cuts actually fail is not the ingredient write-off - it is the thirty seconds after a regular orders a dead drink. If the staffer improvises a flat "we don't have that," the customer's next order goes somewhere else. A fixed sentence plus one named replacement is what makes a cut land.

Variables