Bubble tea new-drink launch SOP - pricing frozen at D-14, a 100% delivery gate at D-7 ("launch short and restock later" and "some stores first" are expressly banned), hands-on staff check at D-3, four fixed data sets collected at D+7; every gate states what to do when it fails and who may delay

You are a marketing & events planner at a bubble tea
chain. Write the new-drink launch execution as an
archivable SOP: a D-14 to D+7 timeline where every
node is a "gate" - with a completion standard, a
failure plan, and a named decider.

Scope: all-store simultaneous launches (including
seasonal rotations). Out of scope: single-store tests.

Timeline (fixed, back-scheduled from launch day D):

D-14, pricing and materials freeze
- Standard: retail and promo prices approved; the
  materials list carries the stock formula - per-cup
  usage x forecast week-one cups x 1.2 safety
  factor; the forecast states its basis (actual
  week-one sales of comparable past launches); a
  forecast with no basis is labeled "a guess" and
  never dressed up as an estimate
- On failure: the whole timeline shifts - the safety
  factor is never what gets cut
- Decider: planning supervisor

D-10, store trial runs and pitch card
- Standard: trial runs in three stores across
  different catchments; pitch card selling points
  <= 2 sentences; trial-run waste booked into the
  launch week's material account
- On failure: change the recipe or the pitch - one
  of the two, no endless rounds
- Decider: the planner plus the trial stores'
  managers

D-7, materials delivery confirmation (the critical
gate)
- Standard: 100% delivery across every store,
  records matching physical stock
- On failure: the delay decision fires. Delaying
  the launch is a legitimate decision; "launch
  short and restock later" and "some stores first"
  are expressly banned - a partial launch hands
  out-of-stock reviews to the launch's loudest
  moment
- Decider: operations manager (a named role - never
  "let's wait and see")

D-3, hands-on staff check
- Standard: at least one person per shift per store
  has independently made 3 cups within the target
  time; new topping labels carry use-by times
- On failure: that shift does not solo-produce on
  day one - shadowing only
- Decider: store manager, spot-checked by the field
  supervisor

D-day, launch and baseline
- Pre-open check: week-one stock in place; shelf
  pricing and the online menu updated together
- Day one records exactly five items: sales, sold-
  out time, overdue-ticket count, sampling
  consumption, the single most-asked customer
  question (fed back into the pitch card)

D+3, midweek check
- HQ checks one thing only: material consumption
  against forecast. Execution drift (pitch not
  delivered, displays not changed) is handed to
  the store-side first-week monitoring process -
  this SOP does not duplicate that check

D+7, retrospective
- Collect exactly four data sets: sales share,
  repeat rate (stores without member identification
  record "cannot measure"), topping waste rate,
  overdue-ticket rate
- The retrospective answers exactly three
  questions: was the material forecast accurate,
  did the pitch card get used, what stock factor
  for next time

Document spec: three A4 pages; one section per gate:
owner / completion standard / failure plan / where
the record is filed. Improvised items (local ads,
cross-promotions) go to an appendix and never touch
the main timeline.

Constraints:
- No "strengthen promotion" or "build buzz" - any
  action without a completion standard is banned
  from the SOP
- Items beyond HQ authority (paid placement) say
  "applied for by [role]," never a wish
- The main timeline contains no "arrange flexibly as
  circumstances allow" - a flexible gate is no gate

Why it is written this way

Back-scheduling from D-day rather than scheduling forward from week one exists because the launch date, once announced, is immovable - every preparation must be pushed back from a deadline. A forward-scheduled plan always discovers "no time left" in the final week and then starts cutting the safety factor and the trial runs, precisely the two cuts that matter most. Of each gate's three parts, the failure plan matters more than the completion standard: a process without a failure plan meets its first problem with "launch anyway and deal with it later."

The D-7 delivery gate is the hardest line in the document because it blocks the most common self-destructive move in launches - "go live short, restock later." The launch's first week carries the most attention and the most damaging reviews; three stores out of stock in one city, and the screenshots of "new drink unavailable" spread at peak volume. A week's delay costs the schedule; a hard launch costs the first impression. "Some stores first" is banned for the same reason: the announcement is national, and customers do not distinguish stores.

The pitch card's two-sentence cap comes from one plain fact: staff in rush hour will not say a third sentence. Five written selling points produce zero spoken ones; two written produce one blurted at the bar. Booking trial-run waste into the launch week's material account keeps the D+7 conclusion "week-one material overrun" from being contaminated by sampling.

The D+7 retrospective collects four fixed data sets and answers exactly three questions to keep the meeting from turning into a brainstorm. The standard derailment is the second hour drifting into "what campaign next" - when what this launch actually needs to bank is three things: forecast accuracy, pitch usage, next stock factor. Handing execution drift to store-side monitoring at D+3 is this SOP's respect for the boundary: HQ owns materials and rhythm, the store owns execution, and each side runs its own checks.

Variables