Inventory counter bean-count convention checklist - how to count sealed bags, opened bags, hopper stock, retained grounds, and steeping cold brew, pinned so week compares to week

You are the inventory count specialist of a coffee shop.
Turn the bean-counting conventions below into a tickable
execution checklist. Every count object states
"convention + tool + record fields," with one goal: the
method is identical every time we count, so weeks stay
comparable. Store parameters stay as 【】 placeholders.

Background (put it in the checklist header): the
biggest source of bean-count variance is not missing
beans but convention drift - estimate the hopper this
week, skip it next, and phantom differences big enough
to drown a real usage anomaly appear. This checklist
pins the conventions.

Count objects and conventions (fixed; execute in
order):
□ Sealed bags: count at the bag label's net weight,
  unopened - spot-check 【2】 bags per batch on the
  scale (packaging included, then subtract tare);
  batches whose actual weight deviates more than
  【2】% from the label get recorded separately and
  escalated
  Tool: scale; record: coffee / batch no. / bag
  count / spot-check result
□ Opened bags: weighed, not gauged - tare the scale,
  pour the beans out, weigh the beans; no visual
  estimation, ever
  Tools: scale + receiving container; record:
  coffee / open date / net weight g
□ Hopper stock: the fixed convention is "grind to
  the line before counting" - run the hopper below
  its minimum graduation mark, then book the
  remaining beans as a fixed 【50】g constant; no
  per-count estimating
  Tool: none (a convention constant); record:
  ground-to-line yes/no - if no, this period's bean
  count is void and gets redone
□ Retained grounds in the grinder: excluded from the
  count (fixed convention) - the amount is small and
  stable week to week, and including it imports
  cleanup variation instead
□ Steeping cold brew: booked as work-in-progress
  from the batching log, on its own line - batch-in
  date and planned serve date noted, never mixed
  into spot stock
  Tool: batching log; record: coffee / grams in /
  date in
□ Same-day waste: test-shot beans (dose × count)
  entered on that day's waste slip, never bundled
  into an end-of-period estimate - waste settled
  daily is what keeps the difference clean

Execution rules:
- Counts happen at a fixed time 【after close /
  before open】, away from service
- Two-person teams: one reads, one records, both
  sign on the spot
- Any convention that can't be executed (hopper not
  ground to line, scale uncalibrated) voids that
  object's count this period - noted in the header,
  redone or left missing; never fudge a number
- The count sheet's columns are fixed: object /
  convention code / value / unit / counter /
  verifier - change the columns and comparability
  breaks

Constraints:
- Every convention must be executable and checkable -
  no "estimate the remainder" or "use judgment"
  entries that vary by person
- This checklist pins conventions only; it changes
  no arithmetic (opening + arrivals - closing =
  usage)

Why it is written this way

A bean-count checklist's first goal is not accuracy but comparability: perfect accuracy is impossible (the hopper is never empty, the grinder never clean), but if the method is identical every week, differences mean something. Let the convention drift - hopper estimated this week, unweighed next - and a few hundred grams of phantom variance buries any real anomaly, at which point the count's early-warning value is gone. Every line in this checklist serves comparability, not precision.

"Grind to the line + fixed constant" is the most valuable line in the file: eyeballed hopper stock means a different number per eyeball, and the count sheet becomes a Rashomon experiment. Pin one action (grind past the minimum graduation mark) and one constant, and the object moves from "estimated" to "booked." The paired rule - not ground to line means the count is void - gives convention violations a price; otherwise a convention is just a suggestion on paper.

Explicitly excluding retained grounds is counterintuitive but the math holds: the amount is small and stable (it rides in the grinding chamber until the next purge), and counting it would require a full purge-and-weigh every time, where purge-to-purge variation imports more error than the grounds are worth. Convention design picks stable-comparable over theoretically-complete - the one line that most needs its rationale explained to a junior, which is why the reason is written into the checklist itself.

Settling waste daily (test shots booked as dose × count from the SOP log) rather than estimating it on count day keeps the difference formula honest: waste is a subtraction term, and an unreliable subtraction makes the whole difference meaningless. The dial-in log already exists, so booking waste from it costs nearly nothing - low cost, hard convention, no reason not to.

Variables