Snack shop inventory count specialist - bulk-bin weighing count SOP: build the tare ledger before counting anything, record gross per bin and let the sheet subtract tare, mixed bins get their own pending-sort lines instead of being merged, everything records in kilograms, and any bin breaching the variance threshold gets reweighed by a second person before it stands

You are an inventory-standards coach for a
snack chain. Turn the bulk-bin weighing count
into an archivable SOP for our next count.

Context: fake variances in the bulk section
come from three sources - estimated tares,
force-merged mixed bins, and pounds logged as
kilograms. Nail down the method first, or
variance analysis is talk.

I will give you: number and sizes of bins,
whether the count runs during trading hours or
closed, staffing, scale precision (minimum
readability) and count, and the three items
that carried last count's largest variances.

The SOP has three sections:

1. Pre-count preparation (complete before
count day)
- Tare ledger: weigh every empty bin (inserts
  included), write the weight on the bin's hang
  tag, and log it - bin number / tare / weigh
  date. Re-weigh every [tare re-weigh cycle]
  months, and immediately upon any bin
  replacement
- Scales: level, zero, then weigh one standard
  object on every scale in the count - any
  scale that disagrees with the others is out
  of the count
- Unit standard: the count sheet records
  kilograms only; scales reading grams convert
  on the spot (1 kg = 1000 g, printed in the
  sheet header); the word "pound" is banned
- Counting-window rules for open-hours counts:
  log the cutoff timestamp, add back bulk items
  sold after cutoff, and freeze restocking and
  sampling in the section during the count

2. Bin-by-bin counting
- Two people per bin: one weighs and reads
  aloud, one records and checks
- Record: bin number / item / gross - net
  weight comes from the sheet subtracting the
  ledger tare; any bin missing from the ledger
  gets weighed empty and entered first, then
  counted
- Mixed bins (customer-scooped cross-
  contamination): never merged into the main
  item - a dedicated "pending sort" line: item
  field carries main + intruder, gross recorded
  as-is, estimated intruder share noted as an
  estimate; the books take the post-sort real
  figures, never the eyeball estimate
- Deep or high bins with unreachable bottoms:
  transfer into a spare bin, weighing before
  and after, cross-checking the pair

3. Verification and booking
- Net differences within twice the scale's
  minimum readability count as weighing noise -
  not chased
- Bins whose net variance (converted to system
  units) exceeds [variance threshold] get a
  second weighing by a different person on a
  different scale; the second result stands,
  and if the two still disagree past threshold,
  both are recorded and escalated
- Pending-sort lines book only after the sort
  completes on real figures; the count stays
  open until then
- Archive: date / window / participants / net
  per bin / variance / reweigh records

Constraints:
- Every step states "what counts as done";
  bins without a ledger tare are never counted
  on an estimated tare - weigh first, count
  second, always
- No "check carefully" / "weigh attentively" -
  the SOP buys accuracy with reweighing and
  two-person actions, not adjectives
- One and a half pages max; append blank
  templates for the tare ledger and the bulk
  count sheet

Why it is written this way

The tare ledger is the foundation and sits first in preparation. Bin self-weights run from a few hundred grams to two kilograms; mis-estimated tares across ten bins can equal an entire real shrink event - and tare is exactly the number people paper over with "roughly eight hundred grams." Making the weighed-and-tagged empty bin an admission condition for counting (weigh it now if the ledger lacks it) removes the largest fake-variance source at the process level.

Mixed bins get pending-sort lines rather than eyeball-share booking - a deliberate trade of honesty against speed. Booking estimated shares closes the count immediately but leaves error nobody can bound; the extra sorting step buys books where every number is a sorted, physical figure. Cross-contamination is routine in snack shops (one crossed scoop makes a mixed bin), and this criterion is what decides whether the count's variance is usable as input to shrink analysis at all. In the same spirit, "within twice the minimum readability counts as noise" sets an honest floor: chasing zero variance means chasing into the noise.

Kilograms-only, printed in the sheet header, with "pound" banned, addresses snack-counting reality: floor talk runs in pounds, system stock runs in kilograms, and the count sheet is the border post between those two worlds - leave the unit loose and 2x data-entry errors appear on schedule. The reweigh-by-a-second-person rule for threshold breaches separates "variance" from "weighed it wrong": every final number on the sheet should survive the question "would it read the same again?" - and only then does downstream shrink analysis stand on anything.

Variables