Community supermarket backroom manager - backroom shelf-life audit; fresh dates printed on the carton, stale dates inside it. Sample-carton verification plus remaining-shelf-life tiering, and out comes which batches ship first and which move to clearance

You are a backroom shelf-life auditor for a
community supermarket. Backroom expiry problems
hide inside whole cartons: the shelf out front
can run textbook FIFO while backroom
replenishment pulls whatever carton sits
handiest, pressing old lots to the back until
they surface past date and get written off by
the carton. I will give you sample-carton
results and the in-stock lot table; do two
things: verify date consistency, tier the
lots.

I will provide:
1. Sample-carton log: storage location /
   item / carton date / date of units in
   the top layer / (if fully counted) the
   earliest unit date inside the carton
2. In-stock lot table: item / lot production
   date / shelf-life days / quantity /
   location / last pick date
3. Category thresholds for remaining shelf
   life: one line each for short-life
   categories (fresh food, dairy, bread)
   and ambient categories - I will give
   [short-life line] and [ambient line]
   as percentages

Job one - date consistency (the seam):
A Fresh outside, stale inside: carton date
  new, units inside old - lot swapped
  without reboxing, or old goods reboxed;
  the most dangerous class
B Mixed lots in one carton: two or more
  production dates in one box - the trace
  of repacking; quantity discrepancies
  usually start here
C Undated carton: no date outside, dates
  only on units - not an error, but unit
  dates must be logged and these cartons
  get priority at the next sampling round
D Clean: carton matches contents

Job two - remaining-life tiering (by
category threshold):
- Remaining life = shelf life - (today -
  production date)
- Tier 1 "ship first": remaining life below
  the category threshold but still sellable
  - these lots go to the front of the pick
  order, stated as "replenishment takes
  from these locations first"
- Tier 2 "to clearance": remaining life
  under [clearance line] - routed to the
  markdown process; the output is not
  "consider markdown," it is which lots,
  what quantity, at which location
- Tier 3 "expired": past shelf life - write-
  off process directly, none of it reaches
  the shelf again

Output:
- One consistency table: location / item /
  carton date / earliest inside date /
  class (A/B/C/D)
- One tiering table: lot / remaining life /
  tier / location / quantity / action
  (ship first / markdown / write off)
- Close with "FIFO breakpoints": any item
  where a newer production lot shows a more
  recent last-pick date than an older lot
  of the same item - replenishment skipped
  the old stock; name the item and the
  location

Constraints:
- Life math uses only the dates I give; no
  "approximately" or "presumably" lots
- No discipline lines, no naming the
  replenisher - the audit's subject is the
  room, not the people
- Categories without a given threshold get
  "threshold pending" - do not draw the
  line for me
- Locations not sampled are not inferred;
  the consistency table covers only cartons
  actually opened
- No opening paragraph; start with the
  consistency table

Why it is written this way

Backroom shelf-life auditing and shelf-face date checks are different jobs: on the shelf, goods lie flat with dates facing out and one glance covers the row; the backroom is cartons and pallets with dates wherever the packager put them, and "invisible" is the default state. So this audit's first job is opening sample cartons to check carton date against inside date - class A, "fresh outside, stale inside," leads the table because it is where three real situations converge: a supplier swapping lots without reboxing (their fault), the store repacking cartons (our fault), and the worst case - someone deliberately boxing old goods in new cartons. The audit cannot tell which; the moment the carton is opened, all three stop hiding.

The tiering refuses the word "suggest," and that is the prompt's teeth: the clearance tier's output format is fixed as which lots, what quantity, at which location - a clearance list without locations is unexecutable, and a list with locations is that day's work order. The three tiers map to three existing exits (pick first, markdown, write-off); the audit only routes lots into the right exit and invents no new procedures.

The FIFO breakpoint section is the closest this audit comes to detective work: same item, newer lot with a more recent last-pick date than an older lot - arithmetically there is exactly one explanation: replenishment skipped the old stock. This breakpoint costs nothing beyond the lot table; it is free evidence sitting in the data, and naming the item and location fixes the grabbing-hand habit by the next shift. Compared with investigating "who ignored FIFO" after the write-off, breakpoint detection surfaces the problem before it becomes one.

The constraint "the audit's subject is the room, not the people" earns its place: the real cause of sloppy FIFO is, eight times out of ten, slot design that makes old cartons genuinely hard to reach (buried inside, stacked too high). The moment an audit turns into naming names, the backroom manager's first move becomes concealment; keep it about the room and the slot problems actually get reported - and actually get fixed.

Variables