The Swap That Is Only an Edit
Two people agree to trade Saturday. Somebody opens the roster and changes a name. The roster is now correct and the trade has left no trace of itself — not who asked, not who agreed, not who approved, and not what it said an hour ago.
The most disputed event in any rostered workplace is not a missed shift or a late arrival. It is a swap — because a swap is the one change that two people make between themselves and a third person has to honour later.
The short version
A roster here records who is working, and it records who last edited the row. It has no concept of a swap as an event, so the trade that produced the edit — the request, the agreement, the approval, and what the shift said before — exists only in whatever the two people can produce on their phones. If your workplace swaps shifts routinely, that is the gap between your roster and your disputes.
What a roster row actually is
One person, one date, one shift. That is the whole shape, and it is the right shape — it is exactly what you need to draw a grid of who is on when, and the grid here is good: seven days across, staff down the side, filterable by department, with leave and public holidays on a companion view so you are not planning against a calendar you cannot see.
It is also worth knowing that the roster keeps the standing pattern and the dated plan as two different things. The pattern says this person normally works earlies. The dated roster says this person is on this Saturday. Systems that collapse those two into one are much worse to work with, and the distinction is why this one can express "normally earlies, but not this week".
Change who is on Saturday and you have edited the row. The row now says the other person. Everything downstream — the grid, the availability board, the count of who is on — is immediately correct.
The roster is right. The reason it is right is nowhere.
What a swap contains that an edit does not
An edit has one participant: whoever made it. A swap has at least two, usually three, and a small amount of history.
| Part of a swap | Where it lives today |
|---|---|
| Who gave the shift up | The previous value of the row, which has been overwritten |
| Who took it | The current value of the row |
| That the two agreed | A message on somebody's phone |
| Why | Unwritten |
| Who approved it | Unwritten — though whoever edited the row is recorded |
| When it was agreed | Unwritten; the edit time is the time somebody typed it |
The three arguments this produces
They are extremely ordinary and every rostered workplace has had all three.
- "I never agreed to that." The roster shows them on Saturday. They say the swap was proposed and never accepted. There is a name on the edit and it belongs to the supervisor, so the record cannot tell you which of the two stories it supports.
- "I gave that shift away three weeks ago." Somebody is marked absent for a shift they had traded. The trade happened, the roster was updated, and then it was updated again — and the intermediate state is gone.
- "He has taken every Saturday off this quarter." A pattern claim, which is the hardest kind to settle. Patterns of swapping are invisible when swaps are not events, because there is nothing to count.
That third one matters more than it looks. Swaps are how an informal seniority takes hold in a workplace — the people with the standing to ask get the shifts they want, and the people without it absorb the rest. It is entirely invisible in a roster that shows only its current state, and it is one of the most reliable sources of resentment on any team.
Why this is common rather than careless
Because a roster is designed to answer one question well: who is working. For that question the current state is not just sufficient, it is the whole answer, and history would be noise. Almost every rostering tool models it exactly this way for exactly that reason.
Swaps break the assumption because they are the one roster change where the route to the current state is itself a fact somebody will need — usually weeks later, usually in a disagreement, and usually with pay attached.
And there is a second-order effect worth naming. Where a swap has no representation, staff stop telling the system about it. They arrange it between themselves, one of them turns up, and the roster is never edited at all — because editing it was never the point of the arrangement. Now the roster is wrong as well as silent, and the attendance record is the only thing that knows what happened.
Running it well anyway
-
Make one channel the record
A single group thread, or a physical book at the supervisor's desk, where a swap is requested and agreed in writing. It does not matter which — it matters that there is exactly one, because two channels is the same as none.
-
Require the giver to ask, not the taker
The person giving a shift up makes the request. This one convention removes most of the "I never agreed" arguments, because consent flows in the direction the dispute will later be about.
-
Update the roster at the moment of approval
Not on the day. A roster edited on Saturday morning is a record of who turned up, which the attendance system already knows.
-
Count swaps by person, once a quarter
From the channel, by hand, on a page. Fifteen minutes and it surfaces the informal seniority before somebody resigns over it.
Three, and the first one is most of the value
The roster, the shift patterns and the availability board all exist. A swap is a transaction that sits on top of them rather than a change to how any of them work.
A swap as a request
The person giving the shift up raises it, the person taking it accepts, and a supervisor approves — with the roster updating on approval rather than by hand. Every part of the trade is then a fact rather than a message on a phone.
A history on the shift itself
What this shift said before, and who changed it, readable from the shift rather than reconstructed from a change log. This is what settles a dispute in thirty seconds instead of an afternoon.
Swaps counted by person and by shift
Who gives shifts up, who takes them, and which shifts are traded away most. The report that makes an informal seniority visible while it is still a conversation.
How it works: you describe the requirement, we return a written scope, timeline and cost, and once agreed it is built into your environment and maintained as part of the product. If your supervisors spend any part of the week adjudicating who agreed to what, the first item is the one to raise.
Talk to us about rosteringThree questions for any rostering system
How does a shift swap get recorded?
What you will hear
"You just change the roster."
How to read it
An honest answer and the common one. The follow-up that matters is whether anything records that it was a swap rather than a reassignment, because those are different acts with different consent behind them.
Can I see what this shift said last week?
What you will hear
Sometimes, through a general change log.
How to read it
A change log is better than nothing and it is not the same as history on the shift. If settling a dispute means an administrator running an audit query, it will not happen for a shift worth four hours.
How many shifts did this person give away last quarter?
What you will hear
Usually no answer at all.
How to read it
The question that finds out whether swaps are events. If they are edits, this is uncountable by construction, and the pattern it would reveal is the one most likely to cost you somebody good.
What AWRA OpsHub does today
- A standing shift pattern and a dated roster as two separate things, so "normally earlies, but not this week" is expressible rather than a workaround.
- A seven-day roster grid of staff against dates with a department filter, and a combined availability board showing leave, shifts and holidays together.
- Shifts with start, end, break and grace minutes, so a shift is a defined thing rather than a label.
- Attendance recorded separately from the plan, with regularisation and verification, so rostered and actually-worked are two answerable questions.
- The last editor recorded on every roster row, so a change has a name against it.
More we can add to your workspace
- A swap as a request, an acceptance and an approval, naming both people and updating the roster when it is approved.
- History on the shift itself, showing what it said before and who changed it, readable without an administrator.
- A reason on a roster change, distinguishing an agreed trade from a reassignment.
- Swap counts by person and by shift, so a pattern across a quarter is visible.
- A rate per person on a shift, so a rostered hour carries a cost as well as a name.
Where we point you to a specialist
- A roster stays a plan and attendance stays a record of what happened, and we would keep those separate. Systems that let the roster be edited after the fact to match attendance produce a document that is always right and never evidence of anything.
- Whether a swap needs a supervisor's approval at all is a policy question about your workplace and, where you have one, your union agreement. We would build the approval step optional and would decline to decide for you whether it is required.
- We would not infer a swap from an attendance record. Two people covering for each other and one person working somebody else's shift without agreement look identical from the door, and a system that guessed between them would be confidently wrong in the cases that matter most.
A swap transaction, shift-level history and swap reporting by person are one piece of work we can scope and quote on.
Ask one supervisor how last month's swaps went
Not whether they happened — how many, between whom, and where that is written down. The answer is usually a phone, and the phone belongs to somebody who may not work for you next year.
Talk to us about shifts and rosters