How do you fill a last-minute shift without ringing round?

Send the open shift to everyone who is free at once and let the first person to accept take it. That turns an hour of phoning into about thirty seconds, and it spreads the work across your whole roster instead of the same four people.

The rest of this is the detail that decides whether it works in practice: who to send it to, what happens when two people accept simultaneously, and what to do when nobody does.

It is seven in the morning. Someone has called in sick for a nine o'clock visit. Every agency that sends people into homes knows this hour, whatever the trade.

Why does ringing round fail?

Three reasons, and only the first is obvious.

It is slow. Each call is two or three minutes including voicemail. Reaching fifteen people takes most of your window, and you spend it on the phone rather than on the client who needs telling.

It is unfair in a way that compounds. You start at the top of the list with the reliable ones. Those people get offered everything, so they either burn out or start declining to protect themselves. Meanwhile the people further down never get offered anything, conclude the work is not there, and drift to another agency. You end up with a roster where the usable part gets smaller every month, and it looks like a recruitment problem when it is a distribution problem.

It hides your real availability. After six calls you have learned six answers. You still do not know whether the other nine were free, so you cannot tell whether you have a staffing shortage or just a slow morning.

Who should the shift actually go to?

Everyone eligible, not everyone. The difference matters, because a broadcast that reaches people who could never take the job trains them to ignore broadcasts.

Eligible usually means four things at once:

• Free at that time, according to availability they maintain themselves rather than something you typed in three months ago.
• Close enough to arrive, which is a distance question and a transport question. Twelve miles is nothing with a car and impossible without one.
• Qualified and currently compliant. This is the one agencies skip under time pressure, and it is exactly when you should not. If someone's clearance lapsed last month, a 7am emergency is the worst possible moment to discover it.
• Not already over their hours, whether that is a working-time limit, a visa condition, or just a person you do not want doing a sixth consecutive day.

Rank what is left by who has worked with that client before. Familiarity is worth more than proximity to most clients, and a returning face turns an emergency into a non-event.

What happens when two people accept at the same second?

This is the question nobody asks until it has happened, and then they never forget it.

Two workers open the notification at the same moment and both tap accept. If the system simply writes "assigned" twice, two people turn up, you pay both or have an argument with one, and the client watches it happen.

The fix is not a faster screen. It is that the acceptance has to be settled in one place, where the second request can be told it lost. Technically this means locking the booking row while the claim is decided, so the two accepts queue up rather than racing. The first gets the shift. The second gets told, immediately and clearly, that it has gone.

Rostavo does this with a database lock on the booking for exactly that reason. It is an unglamorous detail that you only notice when it is missing, and it is worth asking any system you are evaluating how they handle it. If the answer is vague, they have not had two people accept at once yet.

What if nobody accepts?

Have a ladder, and decide it in advance rather than at 7am.

1. Broadcast to everyone eligible. Give it ten or fifteen minutes.
2. Widen the radius or relax the preference, not the compliance requirement. A worker who has not been to this client before is a compromise. A worker whose clearance has expired is not.
3. Offer a premium, if your rates allow it and you have decided the threshold beforehand. Deciding it in the moment means you will either overpay or refuse something you should have paid.
4. Call the client. This is the step agencies delay and should not. A client told at 7:30 that cover is proving hard, with an honest option of a later start or a different person, is far more forgiving than a client told at 8:55 that nobody is coming.
5. Record why it went unfilled. Unfilled shifts are the single most useful recruitment data you have. Three months of "no one free on Thursday mornings in the north of the city" is a hiring brief.

How do you know the cover actually arrived?

Filling the shift and knowing it happened are different problems, and the second one bites at the end of the month.

A last-minute swap creates three records that have to agree: who was scheduled, who actually attended, and who gets paid. If attendance is a text message that says "all done", those three will disagree, and you will find out when a client queries an invoice for a visit that a different person covered.

Check-in from the worker's phone, with the time recorded, resolves it. The hours that go to payroll and the hours that go on the invoice both come from the same recorded attendance, so a swap does not become a billing argument four weeks later.

Does any of this reduce cancellations?

Not directly. People still get ill. What it changes is what a cancellation costs you.

The expensive part of a sick call was never the shift. It was the hour of coordinator time, the client's confidence, and the slow erosion of a roster where only a handful of people ever get asked. Broadcasting addresses all three, and the roster effect is the one that compounds in your favour over a year.

What genuinely reduces cancellations is more ordinary: availability that workers keep current because updating it is easy, rotas published far enough ahead that people can plan, and noticing when someone's cancellation rate changes rather than discovering it after the fourth one.

Does this work outside home care?

The pattern is identical anywhere you send a named person to a specific address at a specific time.

A cleaning company filling a Tuesday morning clean, a pet-care service covering a midday dog walk, a nanny agency finding emergency childcare, a household staffing agency covering a shift: same problem, same mechanics, different vocabulary. The only thing that changes is what counts as "qualified", which is a setting rather than a different product.

See how it works for home care, cleaning, pet care, nanny agencies or household staffing.

A note on what Rostavo does today: open shifts go out by email and by push notification to workers who have installed the app on their phone. SMS broadcasting is on the roadmap and is not built yet, and we would rather say so here than let you find out after you have moved your data across.