Who May Correct a Kiosk Entry, and When to Escalate

Corrections in EDGEBIC by User Solutions sit on a four-rung ladder, and the rung is decided by how far the mistake has traveled rather than by who noticed it. Getting shop floor correction authority right is mostly a matter of teaching the ladder once, because each rung is cheap and the next one up is always available.

The design principle underneath is worth stating first: nothing on any rung erases anything. Corrections are added to the record, kept alongside the original, and the daily totals the plan reads are rebuilt from the corrected values. So a mistake is never a crisis. It is a thing to fix at the lowest rung that can reach it.

The Four Rungs

Read the "reaches" column and the boundary becomes obvious. An operator's authority covers what is happening now and what happened today at day level. A supervisor's authority covers specific punches that have closed. A planner's authority covers the dates and the job's state, which are the fields other calculations lean on.

Rung One: The Ten-Second Undo

After each piece tap an undo control appears for roughly ten seconds and takes the last count back. It exists for exactly one mistake, the double tap, and it handles it without a conversation.

The interesting detail is that the undo is itself logged, so a supervisor reviewing the history sees the tap and the retraction rather than a silently corrected number. That is a small thing that makes the whole ladder trustworthy: even the most trivial rung leaves a trace. See the ten-second piece undo at the kiosk.

Teach operators to reach for it immediately and not to worry when the window passes, because rung two is right there.

Rung Two: The Current Run and the Day

The operator's own authority extends further than most shops assume, and using it is always better than waiting for someone else.

Counts for the current run can be replaced outright through Edit Counts, which opens a dialog for the good and scrap totals. If the defect category changes mid-run, the reason can be changed rather than left wrong. See good, scrap, and rework counts explained.

Day-level entry sits on the same rung. Manual Entry opens the same day grid the planner uses, with one row per day showing scheduled hours and pieces beside the actual figures, and the auto-calculation option that derives the side you did not type from the operation's rate. It is available from the idle screen and the running screen, so a shift that ended without tap-as-you-go punching can still be entered before anybody goes home. See what the kiosk does when the network drops for the other situation where this is the right tool.

What rung two cannot do is reach a specific punch from a previous day. That is the line, and it is drawn deliberately.

Rung Three: A Supervisor and a Closed Punch

Closed punches are corrected by a supervisor in the history drawer, and every correction requires a reason.

The drawer shows every punch and every edit with who and when, filterable by run, setup, down and idle time, or edits. So the supervisor is not guessing: they can see the punch that ran from Friday afternoon to Monday morning, or the pause tagged as waiting that was really a broken spindle, and correct it in place with an explanation attached. See how a supervisor corrects a kiosk punch and reading the kiosk punch history drawer.

Two mistakes belong on this rung specifically, and neither is reachable from below.

A punch left open past the end of work inflates the phase's elapsed record. Paused and idle time never enters the job's production hours, so the daily figures may look right, but the punch itself is a false statement about when work stopped and it should be corrected.

A wrong pause category is worth correcting even though it costs no hours. The four categories are the basis of every downtime analysis you will do, and a machine failure filed under waiting sends maintenance nowhere. Correcting it is a two-minute act with a long payoff.

Rung Four: Dates and Job State

The planner's rung is not "bigger corrections." It is a different class of field.

Actual dates behave differently from hours. An actual start is written once, so a wrong one has to be cleared and re-set deliberately rather than being overwritten by a later punch: see why an operation's actual start is written once. An actual end is validated on the way in, and a save is refused if the end is not after the start, because an inverted pair is always a data entry error. Clearing an end returns the operation to in-progress so remaining work re-plans.

Previously reported daily values are also planner territory in practice, because changing one asks for a reason and an empty reason cancels the save. That reason lands in the logging history with the old value, the new value, and the author. See logging actuals from the planner.

Job state sits highest of all. Completing a job locks actuals entry and marks it done for reports and for the next scheduling run; reopening unlocks entry and returns its operations to the pool the scheduler considers. That is a plan decision as much as a data decision, which is why it should not be reached for when a single figure is wrong.

The One Row Nobody Should Correct

There is a fifth case that looks like a correction and is not, and it wastes more supervisor time than any genuine mistake on the ladder.

A routing step running on parallel work centers appears as two rows: the primary and its sibling. Only the primary is ever logged. The sibling's dates mirror the primary one to one and its hours scale by the configured factor, automatically, on every save. So a supervisor who spots a wrong figure on the sibling and corrects it will watch the correction disappear the next time anybody saves anything on that job, because the sibling is rebuilt from the primary by design.

The rule is short: always correct the primary row and the sibling follows. If the sibling looks wrong while the primary looks right, the suspect is the factor on the routing rather than the actuals, and that is a planner conversation. See why a parallel step's actuals follow the primary.

The Escalation Rule

Give the floor a rule short enough to remember under pressure.

Fix it now if it is now. Undo the tap, replace the counts, enter the day. No permission needed, no conversation needed.

Tell the supervisor if the punch has closed. Say which job, which machine, roughly what time, and what actually happened. That is enough for the drawer.

Tell the planner if it is a date or the job's status. A start that is wrong, an end that should not have been stamped, a job marked complete too early.

The rule works because each step is genuinely cheaper than escalating, so nobody has an incentive to sit on a mistake and hope. The failure mode a shop should watch for is the opposite of hoarding authority: an operator who assumes every correction needs a supervisor stops reporting when the supervisor is busy, and a shift goes unrecorded to avoid getting a number wrong. An unpunched shift is worse than a punched shift with a fixable error, every time.

Reasons Are the Product, Not the Paperwork

Every rung above the undo window records something, and rungs three and four require a reason. The quality of those reasons decides whether the history is useful in six months.

The difference is concrete. "Fix" tells the next person nothing. "Operator tapped complete an hour early, real end 15:40" tells them everything: what was wrong, what the right value is, and that the operator was not being careless. When a customer or an auditor asks about a job, the reasons are what turn a list of adjustments into an account of what happened.

Two habits carry it. Write the reason for someone who was not there, and name the real cause rather than the symptom. Neither takes longer than a vague reason, and only one of them is worth keeping.

The takeaway

Correction authority follows the mistake, not the org chart: the undo window and the day grid belong to the operator, closed punches belong to a supervisor with a reason, and dates and job state belong to the planner. Nothing is ever erased, so the goal is speed at the lowest reachable rung rather than caution. Watch your correction volume as a signal about the floor process, and treat the reason field as the part that gives the record its value. Explore the platform at EDGEBIC, see the upgrade path in RMDB to EDGEBIC, and read on in how a supervisor corrects a kiosk punch and who should log actuals, the operator or the planner.