Skip to content

Part 2 of 7 · Tip pool splitter series ~5 min read

How the pool and hours are collected

Two of the three inputs are easy and one is not, and pretending otherwise is how tip systems lose credibility. Card tips are a figure in a report. Hours are in the timesheet. Cash is somebody telling you a number, and the system should say so.

Key takeaways

  • Card tips come from the card or till report, and are reconciled to the banking.
  • Cash tips are a declaration: a number, a person, and a timestamp.
  • Hours come from whatever pays people, so the two can never disagree.
  • A period cannot be split until all three inputs are present.
  • A late correction reopens the period rather than adjusting the next one.

Three inputs, three assurance levels

Three tip pool inputs converging on a complete periodThree boxes stacked on the left. Card tips, from the till or card report, labelled measured. Cash tips, declared at close, labelled declared. And Hours, from whatever source pays people, labelled measured. All three converge on A complete period, which requires all three present or there is no split. Below it, connected by a downward arrow, is the Splitter, applying the agreed rule. A note says the middle lane is a person's statement and the record says whose.Card tipstill or card reportmeasuredCash tipsdeclared at closedeclaredHoursfrom payroll's sourcemeasuredA complete periodall three present,or no splitSplitterthe agreed ruleThe middle lane is a person's statement, and the record says whose.
Fig 1. The three inputs and where each comes from. The declared lane is labelled differently on purpose: it is the only one nobody can independently verify.
  • Machine learning
  • Management
  • Analytics
  • People

Card tips

A figure from the card processor or the till, for the period, and it should be reconciled to what actually reached the bank. That reconciliation is the one control worth having on this input, because a card tip figure that does not match the banking usually means a refund or a chargeback that reduced the pool after the fact.

Service charge, where it exists, is separate from card tips and frequently has different rules attached. It is a separate input with its own line in the rule sheet, and conflating the two is the most common cause of a genuine dispute rather than a misunderstanding.

Cash tips are a declaration

Somebody counts what is in the jar and says a number. There is no way to verify it and the system should not pretend there is. So the record is explicit: the amount, who declared it, and when.

Making the declarer visible is the whole control, and it is a mild one deliberately. The alternative — a cash count with a witness and a signature every night — is proportionate in some businesses and heavy-handed in most. Recording who said what, and publishing it with the split, is usually enough.

Hours come from payroll’s source

Whatever pays people is what the split uses. If hours come from the rota and pay comes from timesheets, the two will differ — somebody stayed late, somebody left early — and a tip share computed on rostered hours when pay is computed on actual hours produces a discrepancy that nobody can explain.

Completeness

How a period is checked for completeness before splittingA vertical chain of four steps entered by a box labelled Period ends, on a date from the sheet. Step one asks whether the card figure is in; if not it exits to Wait and chase at twenty-four hours. Step two asks whether cash has been declared; if not it exits to Ask whoever closed, by name. Step three asks whether the hours are final rather than a draft timesheet; if not it exits to Wait for payroll, because a draft is never split. Step four splits and publishes the same day. A note says splitting a draft timesheet produces a share that changes, which is worse than a delay.AWS ACCOUNTPeriod endsa date, from the sheetCard figure in?Wait, and chaseat 24 hoursnoCash declared?Ask whoever closedby namenoHours final?not a draft timesheetWait for payrollnever split a draftnoSplit and publishsame daySplitting a draft timesheet produces a share that changes, which is worse than a delay.
Fig 2. Why a period waits for all three inputs. A published share that later changes does more damage than a share published two days late.
  • App integration
  • Security & identity
  • Management
  • Analytics
  • People

Late corrections

A card refund lands after the split. A timesheet is corrected. Somebody remembers a cash tip that was not declared. All three happen, and the wrong response is to adjust the next period, because that produces a week where everybody’s share is slightly off for a reason buried in a previous week.

The right response is to reopen the period, recompute, and publish a correction that says what changed and why. It is more work to explain and much easier to check, and since the whole system exists to be checkable, that is the trade to make.

Next: the arithmetic itself, and the rule sheet that drives it.

All posts