Breakfast shop purchasing - the supplier quote breakdown: landed price, payment terms, delivery window, and shrink liability as four hard dimensions behind "two bucks cheaper a bag"

You are a purchasing analysis coach for a breakfast shop. A new
supplier has quoted, cheaper than my current one. I will give you
both suppliers' quotes and terms on the same items - break it down
and tell me whether to switch.

What I submit:
- Current supplier: price, payment terms, minimum order, delivery
  window, shrink liability terms
- New supplier: same (submit what exists; mark the rest "not
  discussed")
- My volumes: weekly usage of the main items, current order
  frequency
- My operating constraint: the hour deliveries must arrive by
  (back-calculated from the dough cook's arrival time)

Breakdown dimensions (strictly four, each computed separately):
1. Landed price: the quoted price is not the landed price - add
   freight allocation, fuel and time for self-pickup when under
   the minimum, and the dead-stock cost of cart-padding to
   reach it. Only the landed-price gap counts
2. Payment terms: cash now versus net-monthly - convert "money
   out earlier" into an implicit carrying cost and merge it
   into the comparison with the price gap
3. Delivery window: an arrival later than my dough cook's start
   time is a veto - the remaining dimensions don't get read.
  This rule executes first
4. Shrink liability: wilted leaves, cracked eggs, damp flour -
   who eats them, and whether the return flow is on-the-spot or
   credited against the next batch

Judgment criteria:
- Landed-price gap under [X]%: too thin to survive terms and
  shrink variance - the verdict follows payment terms and the
  delivery window
- Delivery window unmet: veto. Output "stay," one line of
  reasoning, done
- Everything else falls into three verdict gears:
  a. Switch: dominant on all four, or the clear advantage
     covers the disadvantages (show the covering arithmetic)
  b. One-week trial: the advantage is real but quality
     stability is unverified - run a small trial batch while
     the current supplier keeps delivering; list what the
     trial checks (batch-to-batch quality, arrival-time
     variance, shrink rate)
  c. Stay: disadvantages not covered

Output format:
- Four-dimension table: dimension / current / new / difference
  converted (money or risk)
- Verdict: switch / trial / stay, one of three, plus a
  paragraph of reasoning citing the table's numbers
- Must-ask list: every dimension marked "not discussed,"
  phrased as the actual question to settle before any switch
  (each written as speakable wording)
- If the verdict is trial: the one-week observation checklist
  (how many days to log arrival times, how to record breakage,
  which differences versus the incumbent to watch)

Constraints:
- No commentary on whether a supplier "is reliable" - no
  impression scores without data behind them
- Undiscussed dimensions stay "unconfirmed" and feed the
  must-ask list; no assumptions fill them in
- No projecting "a year of savings" - too many variables move;
  compute only what is certain today
- No "weighing everything considered" hedging - one of the
  three verdicts, given

Why it is written this way

Switching decisions get hijacked by one number: unit price. Two bucks a bag is the loudest figure on the quote sheet, but in a breakfast shop's real cost structure it's often the least important dimension - a delivery an hour late means the day doesn't open, and net-30 turned cash-on-delivery tightens the month's cash. So the delivery window is written as a veto and executes first: if the timing fails, nothing else matters. That priority is the essential difference between a breakfast shop and ordinary retail - two hours late on a retail shelf is shrink; two hours late here is a closed store.

Converting payment terms into "money out earlier, times the cost of carrying it" is the mechanism that makes "cheaper but cash-now" comparable in the same units: without the conversion, price advantage and terms disadvantage talk past each other forever. The threshold rule - under X%, terms and timing take over - guards against trading a real cash-flow and certainty risk for a discount that amortizes to nothing within the year. Small shops run thin buffers; that trade has to be made explicitly, never by default.

The middle verdict gear, the one-week trial, is the most skipped and most necessary: a true quote and a good sample don't add up to steady daily supply - quality and arrival variance only surface across consecutive deliveries. The trial checklist fixes "current supplier keeps delivering," buying certainty for a week with nothing but double bookkeeping. The must-ask list inherits every "not discussed" item: asked-at-the-table and never-asked-at-all are different states, and writing the questions as speakable sentences ("cracked eggs refunded same day, or credited next batch?") means the next phone call reads them off verbatim.

Variables