Convenience store cleaning supervisor - outsourced cleaning acceptance audit: "never done" and "done but failed" are forced into separate columns, checkpoints tiered by food-safety exposure with one-tier veto power, two consecutive failures point at the vendor's schedule and tools rather than the cleaner's attitude, and fee deductions cite only clauses that actually exist in the contract

You are a convenience store cleaning acceptance
consultant. I am the cleaning supervisor covering
[number of stores] stores; cleaning is outsourced
to a per-visit contractor with checkpoints and
frequencies written into the contract. Turn my
spot-check notes from this round into an
acceptance audit result.

Input I will provide:
- The contracted checkpoint/frequency table:
  checkpoint / frequency, copied verbatim from
  the contract
- This round's spot-check notes: checkpoint /
  time of visit / what I saw on site (photos
  described too, if I have them)
- The redo and make-up list from the previous
  audit

First, tier the checkpoints (the tier decides the
consequence of failing):
- A, food-safety exposure (veto tier): floor
  around the fresh-food counter and hot-food
  island, microwave interior and exterior,
  self-service touch surfaces. If any of these
  fails, the whole acceptance fails no matter
  how good everything else looks
- B, customer-visible: entrance glass and mat,
  visible shelf faces, dine-in tables, shopping
  baskets
- C, not customer-visible: backroom floor,
  trash holding area, shelf back panels and
  baseboards

Verdicts go in two columns and may not be merged:
- "Never done": the contracted visit did not
  happen, or the contractor's log contradicts
  what I found on site - the remedy is a make-up
  visit
- "Done but failed": traces of work exist but
  the pass bar wasn't met - the remedy is a redo
  plus re-inspection

Pass criteria, fixed per checkpoint (criteria I
supply take priority; where I gave none, draft
one per checkpoint in the style below and mark it
"pending my confirmation"):
- Microwave interior: no grease, no food residue,
  no odor, nothing diggable out of the tray
  seam
- Entrance mat: lift the mat - no grit or water
  under it or on the floor beneath
- Shelf baseboard: wipe once with a dry tissue -
  no visible dust line

Output:
- Table 1, spot-check verdicts: checkpoint /
  tier / contracted frequency / condition found /
  verdict (never done / done but failed / pass /
  not inspected) / evidence (photo or note ID)
- Table 2, remedies: never-done items -> make-up
  visit with deadline; done-but-failed items ->
  redo plus re-inspection method (what the
  re-inspection checks, and against which bar)
- Table 3, repeat offenders: cross-referenced
  against last round's list - checkpoints failing
  two audits in a row. This table points at the
  vendor's scheduling or tooling; the remedy is a
  meeting with the contractor's manager, not a
  penalty on the cleaner who was on shift
- Table 4, fee deductions: output only if I have
  provided the contract's quality clause verbatim,
  each line mapped to a clause number; if the
  contract is silent, write "no contractual basis
  - this round claims redos only"

Prohibitions:
- Checkpoints I did not inspect are marked "not
  inspected" - never guess their condition
- Do not invent penalties, deduction rates, or
  liquidated damages the contract does not contain
- No "strengthen cleaning supervision" or "maintain
  communication with the vendor" - nothing that
  cannot itself be inspected
- No recommendations of detergent or tool brands
- When a tier-A checkpoint fails, no hedging like
  "broadly acceptable" or "fine overall" - the
  verdict is pass or fail, nothing between
- No polite preamble; start with Table 1

Why it is written this way

"Never done" and "done but failed" are forced into separate columns because they mean entirely different things to a per-visit contractor: the first is a scheduling and honesty problem - under per-visit billing, missed visits are directly claimable - while the second is a quality and tooling problem: the person came and worked, and the result fell short. Merge them into one "failed" list and month-end billing can neither count the make-up visits owed nor specify what needs fixing. Splitting them also protects the cleaner on shift: a dirty microwave interior is usually the wrong detergent or not enough minutes on the clock, not laziness - a blanket "failed" pushes both causes onto the person holding the rag.

The three tiers with a veto at tier A put weight into the sheet. The classic failure of cleaning acceptance is averaging: backroom floor mopped to a shine, microwave interior crusted with old food - item-by-item scoring calls that a ninety. But the customer heating a breakfast bun stands with their face next to that microwave, and the residue sits right against food. Without weights, an outsourced vendor spends its limited hours where the points are visible; the veto forces hours back onto food-safety-adjacent surfaces. That single rule is the lever the whole table exists to pull.

Repeat offenders get their own table because two consecutive failures change who the conversation belongs to. One failure can be a person's bad day; two in a row is a system - not enough hours priced in, the wrong chemicals, nobody ever trained on degreasing tile grout. Dressing down the cleaner on shift is the most common and least effective response: next month the same checkpoint fails again. This table exists to move the meeting up one level.

Fee deductions may cite only real clause numbers because the model will happily invent penalty terms - "deduct twenty percent of the per-visit fee" sounds perfectly reasonable the moment it appears. Written into an acceptance sheet and read out at the monthly billing meeting, an invented clause destroys credibility in one sentence. So Table 4 is produced only from the user's verbatim contract text, and where the contract is silent the sheet honestly claims redos only - a redo claim depends on no contract detail and always holds up.

Variables