Bakery inventory count specialist - the count-scope checklist: six layers counted in full, tare off packaged goods, frozen weighed frozen

You are a count process coach for a bakery. Build me a count-scope
checklist for the count specialist to run every week right after
close. The goal: lock every convention once, then copy the same
sheet weekly - no convention decisions on the count floor. Stock
sits in six layers: dry storage, freezer, cooler, work-in-process,
counter product, packaging.

Four blocks:

One - all six layers counted (one missed spot in any layer scraps
the whole count):
- Dry storage: sealed bags and cases counted by unit; opened bulk
  flour and sugar weighed separately, never merged into the sealed
  line
- Freezer: butter, frozen dough, fillings - each item weighed
- Cooler: eggs, dairy, opened items moved to chill, dough in proof
  - each item weighed
- Work-in-process: proofing, in the oven, cooling - recorded at
  input weight, reconstructed from the recipe and the requisition
  record, never a rough headcount; an estimated number in the sheet
  is a fake number
- Counter: fresh-baked leftovers, prepacked product, markdown
  section - three separate lines
- Packaging: boxes, bags, stickers, ribbon, counted by unit - the
  layer most often skipped, and every line of it is money already
  spent

Two - weighing conventions (everything actually weighed; the count's
credibility comes from that):
- Packaged goods: look up the tare table and subtract; any package
  type missing from the table gets an empty package weighed on the
  spot and added
- Cases and bags booked at label weight: open one and weigh it each
  count to verify the label is trustworthy; if the sample misses,
  that item gets every case opened and weighed this time
- Frozen goods weighed in frozen state, never thawed and converted
  - a thaw conversion is an estimate, a frozen-state reading is an
  actual weighing
- Eggs counted by piece, never weighed
- Units fixed in each layer's header (kg/g, pieces); one column
  never carries two units

Three - recording conventions:
- Six layers in separate columns, never summed - dry goods turn
  weekly, cooler and counter daily, and a grand total supports no
  decision
- Opened items get their own line tied to the opened-on label
  (date, who opened it)
- Short-dated items flagged in the notes column; how many days out
  counts as short is mine to set by category

Four - book-to-floor check:
- Theoretical stock = opening + purchases - production requisitions
  - sales relief - shrink; compute per line and compare against the
  count
- Variance inside my tolerance line: log and file it, no action
  this period
- Variance over the line: no guessing at causes on the count floor;
  the whole sheet moves to the variance-review process

Format: four checkable blocks, plus one fixed-header sheet per
layer (item / unit / counted / theoretical / variance / notes) for
weekly photocopying.

Constraints:
- No invented "reasonable shrink rate" or "normal variance band" -
  the tolerance line is mine, set from my own recent variance
  history, never copied
- No recommending a close-the-shop full count - a cycle count that
  survives beats it; a precise convention nobody keeps past two
  months is worth nothing
- Design the checklist to finish inside half an hour after close;
  anything that overruns gets cut or simplified, not staffed up

Why it is written this way

The six layers and the scrap-the-count rule exist because bakery stock never sits in one place: dry goods in the store room, dough in the proofer, product on the counter, packaging on a shelf in the corner. What gets missed is rarely the valuable stock - packaging and opened bulk goods do not look like inventory, and a count that skips a layer does not under-report, it lies: the book-to-floor check books every missed item as a loss, and next period's investigation chases ghosts. Scrapping the count is harsh on purpose; it is the only enforcement that makes anyone walk the layer list, and saying "count everything" has never worked.

The weighing block is built on one principle: every number is actually weighed. A thaw-back conversion drags a yield factor in, and a factor is an estimate - verifying the books against estimates resets the count's credibility to zero. Weighing frozen while frozen costs cold hands and nothing else. The single-case verification is the middle path: opening every case is too slow, trusting labels forever is a standing bet against short-weight deliveries; one case per count, and only a failed sample escalates to opening everything. Eggs by piece because egg weight drifts with size grade - weighed, they produce pure noise; counted, a one-egg gap is a real gap.

Layers stay in separate columns because their clock speeds differ: dry goods move weekly, cooler and counter daily. A summed figure answers neither "does the flour last to next week" nor "was today's shrink abnormal"; it just buries small variances inside a big number. Tying opened items to their opened-on label welds the count to date-code control - the sheet ends up saying not just how much is left, but what should have been used up already. The short-date flag lives in the notes column rather than on a separate sheet, again to protect the half-hour cap.

Both constraints aim at sustainability. A close-down full count looks more accurate and dies of its own weight - in a bake-today-sell-today shop the theoretical identity holds every single day, so frequent light counts catch far more than rare precise ones. The half-hour cap disciplines the checklist itself: a convention's value equals the number of weeks it survives, and a count that runs an hour in week one is dead by week three. And over-line variance never gets diagnosed on the spot, because at ten at night the crew can record but cannot investigate - a guessed cause written down closes the question and buries the real one.

Variables