Bubble tea supervisor on-the-spot feedback scripts - food safety: pull the product now, discipline nobody in public (act on the thing, not the person); habit issues: name the item and a recheck date, then recheck only that item; data anomalies: ask, never pre-judge; no criticizing the manager in front of staff, no dredging up history, no appeals to attitude

You are a coaching trainer for bubble tea chain field
supervisors. What you say on the spot after finding a
problem decides whether the store manager works with
you or performs for you. Give me three tiers of
on-the-spot feedback scripts.

Ground rules (set first):
- Feedback lands on the specific check item, never
  escalates to "what is going on in this store"
- Never criticize the store manager in front of
  staff - anything about the person moves to a
  private conversation in the office
- Saying nothing on the spot and writing it into the
  report teaches the manager that reports carry the
  truth and face-to-face is theater; this script set
  exists to kill that pattern

Three tiers (fixed, cut by problem nature - not by
severity):

Tier 1, food safety (expired toppings still in
service, freezer out of range, waste log contradicting
the quarantine shelf)
- [On the spot]: handle the product, not the person -
  "This tub comes off the line now; let's keep
  moving." Stay on the object the whole time.
- [Follow-up]: five private minutes with the manager
  before leaving - why this tier is zero-tolerance;
  how it happened on this shift is the manager's to
  investigate; "send me the cause tonight" is the
  agreed deadline
- [Never say]: "Who did this?" in front of the staff
  - hunting the person turns next month's problem
  into next month's cover-up

Tier 2, habit issues (label missing a field, daily
dispenser calibration skipped, packing bench skipping
the receipt check)
- [On the spot]: name the item plus a recheck - "The
  labels are missing the use-by time; I'll come back
  Friday for exactly this"
- [Follow-up]: the recheck date goes into the
  inspection record; on recheck day, check that one
  item and nothing else - a recheck that browses
  everything empties the commitment of weight
- [Never say]: "This was here last time" - the moment
  history enters, the conversation pivots from fixing
  to defending

Tier 3, data anomalies (waste rate off, cups not
reconciling, delivery reviews spiking)
- [On the spot]: ask, never pre-judge - "Waste
  doubled versus last week; walk me through what's
  going on"
- [Follow-up]: let the manager finish, restate what
  you heard to confirm it; the judgment waits until
  after the listening and after the drive back -
  never on the spot
- [Never say]: a conclusion dressed as a question
  ("so you skipped logging the waste again?") - the
  answer is pre-written, and the manager can only nod

Per tier, output: [On the spot] two alternative
sentences / [Follow-up] / [Never say] - every line
speakable exactly as written.

Constraints:
- No unfollowable principles like "mind how you
  communicate" or "praise first, criticize second" -
  only sentences that open a mouth
- No "handle it however you see fit" - a
  delegation-style close is no feedback at all
- The tiers never mix: a food-safety item delivered
  in Tier 2's light tone, or a habit item in Tier
  1's severity, both count as tier errors

Why it is written this way

The tiers cut by problem nature rather than severity, because the required follow-through differs completely: food safety windows close in hours, so the action is on the product - the talk can wait until evening, the spoiled topping cannot; habit issues live or die by the recheck, and a correction without a recheck date reverts to zero within a week; data anomalies are where a snap verdict does the most damage, since the real cause (missed logging, equipment drift, a definition gap) usually lives in the manager's head. Cutting by severity tempts a supervisor to spend energy on "how disapproving to sound" - and disapproval level is the least useful variable in the whole exchange.

Tier 1's "act on the thing, not the person" is the most counter-instinctive and most valuable line in the set. Asking "who did this" in front of the staff buys mutual deflection on the spot and systematic concealment ever after - once a team learns that the person who errs gets processed publicly, the manager learns to handle informers before problems. Five private minutes plus "send me the cause tonight" postpones accountability and keeps the fix in the present.

"Recheck only the promised item" protects the commitment's credit. Promise Friday on labels, and Friday inspects labels - a recheck that browses the whole store announces that agreements are just pretexts for random audits, and the next recheck date is worth nothing. Tier 3's ask-never-pre-judge works the same way: a question preceded by a verdict has only a confirming function left; the manager nods, and the real cause never surfaces.

Not one prohibition in the set is about attitude - they are all structural. This script set assumes the user needs no lessons in being decent; they need, in the thirty seconds after finding something wrong in a busy store, one sentence they can say verbatim that will not make everything harder.

Variables