Trainer SOP-change retraining procedure - one-page diff briefs, the schedule blocks untrained staff from changed steps, obsolete copies stamped and archived, first-article check on day one

You are the trainer at a central kitchen. Recipe, process,
label, and packaging changes arrive almost monthly. I need
an archivable "SOP change retraining" procedure that solves
one problem: the document changed, the hands did not, and
the floor still runs the old version.

Process (starts at T-72 hours before effective date; every
step carries a deadline and a deliverable):

T-72h  Change-order review
- Input: the change order (what changed / affected
  processes / effective date)
- Deliverable: the affected-positions list - which
  stations, which shifts must be retrained. Positions
  not on the list are left alone; no all-hands training

T-72h to T-24h  Diff-brief build
- Train the diff only: old vs. new line by line,
  producing a one-page "diff brief" - what changed, why
  (one sentence), what did not change (so nobody
  improvises edits to adjacent steps)
- One A4 page maximum. Over one page, the change goes
  back to the originator for phasing - a change too big
  for one page is too big for the training system to
  absorb in one drop

T-24h to effective date  Layered coverage
- On shift: 10 minutes at the pre-shift meeting, diff
  brief only
- On leave / night shift: retrain at the first pre-shift
  meeting on return. Until then, no work on the changed
  step; a signed-off worker covers. This goes into the
  scheduling rules, not into verbal instruction
- Each person signs: trained on and understood the diff
  brief

Effective date  First-article check
- The line lead verifies the first piece of the changed
  item against the new version, item by item, and signs
- First article fails: stop the line, retrain, and check
  whether the diff brief missed a critical difference

Effective date  Obsolete-version recovery
- Paper copies collected and stamped VOID, archived (not
  destroyed - traceability); electronic copies pulled
  from floor terminals
- Count recovered against the historical issue log;
  every missing copy is traced to its location

Archive package (six documents; incomplete packages are
not archived): change order / affected-positions list /
diff brief / sign-off sheet / first-article check /
obsolete-copy recovery log

Standards:
- What happens to unsigned staff (blocked from the
  changed step) matters more than the sign-off
  percentage - making the rate a KPI breeds proxy
  signatures, and a forged signature is worse than an
  untrained hand
- The affected-positions list follows the change order;
  the trainer neither widens nor narrows it. Doubts go
  into the archive package as notes, confirmed in
  writing by the change originator

Constraints:
- No "continuously follow up on training effectiveness" -
  the deadline on each step is its acceptance criterion
- The diff brief carries the change order's content
  only; no drive-by edits of "while we're at it" items

Why it is written this way

Training the diff, not the full document, is the first watershed between retraining that works and retraining that doesn't. Run the change session as a full-text recital and it cannot finish in twenty minutes; what the floor takes home is "another meeting." A one-page diff makes "what changed" the only payload. The dedicated "what did not change" column prevents improvisation: after a change, the most common incident is not mishandling the new step but staff assuming "everything changed this time" and quietly adjusting adjacent unchanged steps. The one-page limit and phased-rollback clause put a gate on the originator - big-bang changes get trimmed to a sane size by their own scheduling cost.

Blocking unsigned staff via the scheduling system, rather than verbal emphasis, is what turns coverage from a discipline problem into a structure. "No changed-step work before retraining," stated verbally, breaks on day two of peak season - somebody has to run the line. Written into rostering, coverage no longer depends on anyone's conscience. The companion standard - sign-off rate is not a KPI - exists because rate targets breed the path of least resistance: proxy signing. A forged signature is strictly worse than an untrained hand, because the untrained worker at least knows it.

Obsolete copies stamped and archived, not shredded, serves traceability, not tidiness. After a change-related incident the first question is "which version was the floor holding that day" - destroy the old version and that question is unanswerable forever, with "execution error" absorbing the blame by default. Reconciling recovered copies against the issue log prevents old versions resurfacing from a drawer; the missing-copy hunt looks pedantic until the day it is the evidence chain.

The six-document archive package converts "we did train them" into "we can prove we trained them." When an audit or an investigation requests the file, producing the full chain from change order to recovery log in one motion moves the trainer from "can't say whether anyone was trained" to "every link has a signature."

Variables