Bubble tea new-product three-day training SOP - spec card swapped on day one, blind taste test gates the bar on day two, rush-hour shadowing on day three, and no part-timer slips through

You are the training lead for a bubble tea chain.
Write a three-day new-product launch training SOP for
store trainers to execute during launch week.
Context: the brand launches roughly monthly; stores
run mixed full-time and part-time schedules; the
historical causes of launch drift are stale spec
cards and part-timers never trained.

Cover three days, each stating who, when, the action,
and the record:

Day 1 - stock arrival day:
- Spec card swap: the new card goes up on arrival;
  where an existing recipe changed, the old card
  comes down the same moment. No sale before the
  card is swapped - this is a hard gate.
- Pre-shift briefing, 10 minutes, exactly three
  things: what the new drink is; how it differs
  from its closest existing drink (ingredients /
  steps / hold time); how to answer "how is it
  different from X."
- One briefing per shift with sign-in; late-arriving
  part-timers get a make-up briefing from the shift
  lead - also signed in.

Day 2 - hands-on and blind test:
- Small-group practice: each person makes [1] cup
  unsupervised; the trainer observes without
  interrupting and logs the step numbers where they
  hesitate.
- Blind test gate: each person tastes three cups
  (new and old recipes mixed); calling at least
  [2] correctly is a pass. A miss means back to
  the spec card on the spot and a retest next day.
- Results logged per person: passed / pending
  retest.

Day 3 - rush-hour shadowing:
- The trainer stands at the bar for [1] peak hour
  and logs only two event types: first-order errors
  on the new drink (by step number) and customer
  questions the staff could not answer (script
  gap).
- At close: the three days of records go to the
  manager; anyone who failed the blind test but
  is scheduled on the new drink today is flagged
  for the manager to resolve the roster.

Hard gates (their own bolded block):
- No sale before the spec card is swapped
- No solo work on the new-drink station before
  passing the blind test
- A part-timer scheduled into zero training
  sessions this week has their new-product orders
  escalated to the manager - "they'll learn it on
  the job" is not a fallback

Constraints:
- Training content covers the new product only.
  No opportunistic refresher of the whole menu -
  cramming a month of material into three days
  trains nothing.
- Customer scripts cover exactly two questions:
  "how is it different from X" and "what if it
  is sold out." No upselling scripts - the
  trainer's deliverable is consistent execution,
  not extra sales.
- No new forms or systems: sign-in and blind-test
  logs go into the existing training notebook.

Why it is written this way

The blind test is the hardest gate and the part that separates this from "teach the recipe once." Making a cup proves you can follow the card; the blind test proves the recipe lives in your head. Launch drift has a signature failure mode - old-recipe habits bleeding into the new drink, the ice spec or tea base from the closest old product applied by muscle memory - and only a blind test exposes that mixing on the spot. Three cups with a two-correct pass bar leaves tolerance for palate fatigue; requiring all three would block a wave of perfectly usable staff from the bar.

"No sale before the card is swapped" is written as a precondition, not a reminder, because the accident already has a script: stock arrives, the register starts selling, and the wall still shows last month's card - that is the first source of launch drift. "Remember to swap cards promptly" has no teeth; "no sale" does, and it makes the manager personally check the wall.

The briefing covers exactly three things to prevent overload. Ten minutes cannot hold a recipe lecture, and staff walk out remembering the first three items anyway; forcing the differences into three lines makes the trainer decide what is actually new about this drink. If it does not fit in three, the launch plan is the problem.

The part-timer gate stands alone because part-timers are the leak in every launch: scattered shifts, and "they'll pick it up on the job" is the standard cover for never scheduling them. Closing that route and escalating their orders to the manager does not punish the part-timer - it makes whoever builds the roster take training sessions seriously next launch.

Variables