The handover problem.

Lede

The most consequential meeting in a hotel happens standing up, lasts a few minutes, appears on no calendar, and is conducted by two people who both want it to be over.

The meeting that is not on the calendar

Every property has handovers, and almost no property has designed them. They are inherited. Somebody at some point started writing a logbook, or started sending a message at the end of shift, or simply started arriving early enough to be told things, and that became the practice.

What makes the handover strange is that it is simultaneously the highest-value information transfer in the building and the one with the least structure around it. A hotel will have a documented procedure for how a bed is made and no documented procedure for how the evening manager tells the night manager what is currently on fire.

State and judgement

A handover carries two kinds of thing, and they behave completely differently.

The first is state: which rooms are out of order, which arrival is delayed, which group is checking out early, which lift engineer is coming back tomorrow. State is factual, listable, and mostly recoverable. If it is not handed over, the arriving person can usually find it, at the cost of some time and a couple of phone calls.

The second is judgement: the guest in 412 is not actually angry about the room, he is angry about how long check-in took, so do not offer him a room move, offer him an apology from a manager. The supplier who is coming at nine has been late twice and will need chasing at half past eight. The housekeeper on the fourth floor is covering a shift she did not want and should not be asked for another favour tonight.

Judgement is not listable. It is not in the PMS and it will never be in the PMS. It exists because a person watched something happen and drew a conclusion, and it is the single most valuable thing a shift produces.

Now the uncomfortable observation: verbal handovers preserve state reasonably well and judgement almost not at all. Not because people are careless, but because state has an obvious prompt — you look at the board and read off what is on it — and judgement has none. Nobody stands there and asks themselves, systematically, what did I conclude today that the next person will need.

The leak on the third floor

Karim hands over to Lena at three in the afternoon.

He tells her about the leak on the third floor, because the leak is state and the leak is on the board. What he does not tell her is that he spoke to the guest in the affected room himself, in the corridor, and told her that a technician would come before dinner. He does not tell her because at the moment of handover it does not feel like information. It feels like something he did. It is finished, in his head, and the head is where it stays.

At twenty past seven the guest comes to the desk and asks Lena where the technician is. Lena knows about the leak. She does not know about the promise. So what follows costs the hotel far more than it looks: she reconstructs, in front of the guest, a conversation she was not part of, while the guest watches her not know.

The guest's conclusion is not “the technician is late”. It is “they did not write it down”. That conclusion is correct, and it is the one that ends up in the review.

Notice what the failure was not. It was not a missing task — engineering had the leak. It was not a communication breakdown — Karim and Lena spoke, properly, and Karim is good at his job. The failure was that a promise made to a guest existed only as a memory in a person who had gone home.

Why better forms do not help

The instinctive fix is a better handover form. Properties try this constantly, and the artefacts are always the same.

Either the form is thorough, in which case filling it in takes long enough that it gets done at the end of a shift by someone who wants to leave, and the quality is what you would expect. Or the form is short, in which case it degenerates within a fortnight into a field containing the words “all quiet”, written by a person who had a perfectly eventful evening.

There is a third outcome, worse than both: the form becomes a report to management. The moment a handover is read by someone who does not work the next shift, it starts being written defensively. It records what makes the outgoing shift look competent rather than what makes the incoming shift effective. Every property that has ever added a handover to a manager's morning reading has watched this happen.

The reason forms fail is structural. A form asks a person to serialise, at the tired end of a shift, information that was generated across the whole of it. Human beings are bad at that, and being bad at it is not a character flaw.

The briefing that assembles itself

The alternative is not to write the handover better. It is to arrange things so that most of it does not need writing at all, because the things that must survive the shift already exist as objects before the handover starts.

That means, concretely:

• The leak is a task with an owner and a due time, so it survives without anyone remembering it.
• The promise to the guest lives in the guest's thread, because Karim answered her there, or because he added a line to it after speaking to her in the corridor — a smaller act than filling in a form, and one performed at the moment the promise was made rather than seven hours later.
• The room is flagged on the cleaning board, so the state is visible rather than recalled.
• The unfinished checklist is unfinished on the screen, not in somebody's conscience.

Once that is true, the briefing changes character. It stops being a transfer of memory and becomes a summary of state changes, and a summary of state changes is something a system can assemble: what opened, what closed, what is overdue, what escalated, which threads have an unanswered guest at the end of them. In Treema that is LORA's job — it takes the shift's tasks, threads and boards and produces the briefing, so that the outgoing manager is editing a draft rather than composing from nothing.

And then the human part gets shorter and better. Karim reads the assembled briefing, and the only thing he needs to add is the part no system could have produced: 412 is not angry about the room. That is one sentence of work, it is the highest-value sentence in the shift, and it is what the handover should have been all along.

A handover is good when the arriving person can start work without asking a question, and great when the questions they would have asked have already been answered by the state of the board.

Most properties are trying to fix the handover by improving the conversation. The conversation is not the problem. The problem is everything the conversation is being asked to carry that should never have been carried by a conversation in the first place.