A Restaurant Week With Two Missed Clock-Outs

Marcus is a line cook at a 14-seat spot that runs a shared tablet by the pass. He taps his name to clock in, taps it again on the way out. Most nights that works fine. During one busy week he taps in and, twice, walks out the back door without tapping out. If nobody catches it, your Friday payroll pays him for hours he did not work.

Here is the whole week, with the two missed clock-outs called out. The scheduled cutoff for a shift here is 9 hours, meaning any open entry that runs past 9 hours gets treated as forgotten.

What the raw week looked like

Marcus clocked in fine every day. On Tuesday and Saturday he forgot to clock out. Everything else is a clean entry.

The four clean days add up to 27.70 hours. Fine. The problem is the two open entries. A naive system leaves an open entry running until someone notices, then either closes it at the current time or asks you to guess. Say you catch Tuesday's on Thursday morning at 9:15 AM. If the entry is closed at that moment, Marcus reads as clocked in from 4:02 PM Tuesday to 9:15 AM Thursday. That is 41.2 hours for one shift. Do that on Saturday too and you have added a second monster entry before you ever open payroll.

What the inflated total would cost you

Left uncorrected, the week reads like this: 27.70 clean hours, plus a 41.2 hour Tuesday phantom, plus a Saturday entry caught Monday at open (say 42 hours). That is roughly 111 hours for a week where Marcus actually worked about 44. At $17 an hour you would be cutting a check for around $1,140 more than you owe, for one cook, in one week. And you would not notice unless you read every row, because the weekly total is one number at the bottom of a report.

This is the trap with any system that treats a missing clock-out as an entry that is still running. The clock does not stop when the person leaves. It stops when a human happens to look, and humans look late.

How auto-close caps the two bad entries

Kangaroo Clock closes a stale open entry at its start time plus the cutoff, never at the current time. That single rule is what keeps a forgotten tap from becoming a payroll error. Read the full definition on the auto-close glossary page, but the mechanics are short: with a 9 hour cutoff, Tuesday's entry starting at 4:02 PM closes at 1:02 AM Wednesday and reports as 9.00 hours. Saturday's entry starting at 2:30 PM closes at 11:30 PM and reports as 9.00 hours. Neither one grows while you sleep.

The corrected week:

That totals 45.70 hours. Still a little high, because Marcus's real Tuesday and Saturday shifts were shorter than 9 hours, but it is bounded. A capped entry can be too high by a shift's slack. It can never be too high by a day and a half. The difference between a 9.00 hour cap and a 41 hour phantom is the difference between an easy fix and a payroll you cannot trust.

Fixing the two capped days before payroll

Auto-close protects the total. It does not read minds. If you know Marcus really left Tuesday at 10:30 PM, edit that entry down from 9.00 to the real 6.47 before you export. The cap buys you the time to fix it calmly instead of discovering a five-figure error at the register on Friday. When the numbers look right, run a clean CSV export with its distinct-worker count and hand it to whoever cuts checks.

Two habits keep this rare. Post a QR sign-in poster by the back door so the last thing Marcus passes is a reminder to tap out. And read the weekly report before payroll, sorted so any two entries that both hit exactly the cutoff jump out. When a day shows precisely 9.00 hours, that is your signal to check whether a real clock-out is missing.

If you want to see the tap-in flow the way Marcus does, open the live kiosk demo, then start a workspace and set your own cutoff. The free tier runs to 500 recorded hours before anyone asks for a card, which is plenty to run a real week and check the math yourself.