The Night The System Was Down

The connection is down, or a service is unavailable, and it is 20:00.

The room does not care. You are open in an hour either way.

What actually stops

Worth knowing in advance, because it is rarely everything:

Taking bookings. But the ones already made are usually knowable if anyone printed or noted them.

Payments. ⚠️ The one that hurts most, and the one with a physical fallback.

Guest history. Who this person is, what they like. Available only in the memory of whoever is on.

Recording. Everything that happens is happening unrecorded, and that is tomorrow's problem.

The half-hour of preparation

Done once, in daylight, and it holds for years.

Print or export tonight's bookings, daily. ⚠️ Yes, on paper. It costs nothing and it is the difference between a manageable night and a guessed one.

Know the payment fallback. ⚠️ Test it once — a backup terminal or method nobody has tried is not a fallback.

A form for recording what happened. Time, table, what was taken, what was promised. Anything, as long as it exists — a blank sheet with the right headings.

One number to call. Who to contact, and who decides whether to open.

During

Tell people it is happening. Staff first, then guests where relevant. ⚠️ "We are writing things down tonight" pre-empts most of the friction.

Write more, not less. Every promise, every adjustment, every unusual thing. It feels excessive and it is not.

Take names. ⚠️ The thing most often lost — without them, nothing from the night can be reconciled or followed up.

Do not improvise on money. A stated fallback, applied consistently, beats several different approaches by several different people.

The part everyone forgets

Putting it back.

⚠️ The outage gets handled and the recovery does not. The system returns, everyone is relieved, and a stack of handwritten notes sits somewhere.

Enter it the next day, not next week. A day later it is transcription; a week later it is interpretation.

Assign it to one person, with the time to do it. ⚠️ It is real work and it needs real hours — expecting it to happen alongside a normal shift is how it does not happen.

Then check what is missing. Some things will be. Better found on Tuesday than at month end.

Afterwards

Write down what did not work. While it is fresh. ⚠️ The fallback that turned out not to exist, the number nobody had, the thing nobody thought of.

Fix one of them this week. One is enough — and it is more than most venues do after an outage.

Three to hold

Whether tonight's bookings exist outside the system. ⚠️ Check today. It is a yes-or-no question and the answer determines how bad the next outage is.

Whether the payment fallback has been tested this year. Untested is the same as absent.

Days from an outage to the records being entered. Rising means it is not being treated as work.

Where the record has to sit

An outage is survivable when the day's essential information exists somewhere that does not depend on the thing that failed, and recoverable when there is one obvious place to put the notes back. Both are decisions made in advance, on a quiet afternoon.

tasteck keeps bookings and takings on records that can be exported before service, so a night can be run from what was printed and reconciled afterwards.

Nothing about the night was unmanageable. What made it hard was that everything had to be invented at 20:00, by people who were also opening.