The Night the System Went Down

The power goes. The connection drops. The screen will not load.

The venue keeps running. Bookings arrive, guests turn up, drivers go out. The only thing that stops is the record.

And the next morning, nobody can say what that night actually took.

What you lose

Bookings     who came, and when
Assignment   who they were with
Settlement   what was taken, and by which method
Dispatch     where a car went, out and back
Stock        what moved

Settlement is the one you can rebuild, because the cash is physically there. Everything else is gone unless somebody wrote it down at the time.

Decide the fallback before you need it

You cannot design a recording method during an outage. It is busy, possibly dark, and the phone is ringing.

Three sheets cover it:

One    Bookings   time / guest / staff / package / payment
Two    Dispatch   time / destination / driver / out / back
Three  Events     time / what / who handled it / outcome

Print them and keep them where the work happens. You cannot print during a power cut.

Three things that never reconcile afterwards

1  Booking count   what was written down but never entered
2  Times           handwriting says "about"
3  Payment method  recorded as cash, was actually card

The first is the biggest and the quietest. Somebody enters the sheets the next day; if it is busy, a few fall out. Those bookings exist in neither your revenue nor your guest history.

Name who declares the entry finished. That one decision removes most of the loss.

The order to re-enter in

1  Enter the bookings — keep the handwritten times, do not "improve" them
2  Attach the guests — create new records where needed
3  Reconcile settlement against the physical cash
4  Record the discrepancy, as a discrepancy

Almost nobody does step four. Absorbing an unexplained difference means next month's difference has nowhere to sit either, and you get a recurring "somehow it never balances."

Record the outage itself

Stopped at    the last moment it was working — not when you noticed
Back at
Scope         everything, or partial (bookings only, and so on)
Told          who you contacted, and what you said

"When we noticed" and "when it stopped" are different times. In a busy room that gap can be half an hour, and that gap is exactly your missing record.

Contacting guests early is the cheap option

While it is down, you do not know whether their booking exists either.

You contact them      costs effort. Costs no trust
You do not            free if nothing goes wrong. One miss is expensive

⚠️ Do not let queued notifications flush all at once on recovery. Messages that piled up during the outage will land together — several to the same guest, confirmations for evenings that already happened.

The system we built

tasteck is a booking and analytics system for night venues, built by people who ran them for sixteen years and grew from ¥200 million to ¥1.2 billion a year — six-fold, by attacking the business with systems rather than by pushing harder on sales.

The honest part: things do go down. We run on AWS in Tokyo, with daily automated database backups retained seven days and point-in-time recovery inside that window. We run single-AZ with no read replica — so an availability-zone failure is a restore-from-backup event, not an automatic failover. We publish no uptime percentage, because we do not run external uptime monitoring that would let us stand behind one. → Security and data

What helps afterwards: bookings can be entered with a specified time, settlement is broken out by line so a discrepancy can sit on its own line, and guests can be attached to visits after the fact.

What it does not do: we supply no paper templates, we do not notify your guests, and there is no mechanism that keeps running on your behalf while we are down.

The output no one else produces

tasteck outputs the maximum you can spend on each marketing channel next month, as an amount in your currency.

A lost night distorts it. Bookings that never got entered drop straight out of the record for whichever channel delivered those guests. One night is noise; once missed entry becomes a habit, channels change places in the ranking. The ceiling is computed from the lifetime value of the guests each source actually delivered — so a channel looks cheaper exactly in proportion to what you failed to record.

Nothing else in the nightlife category produces that figure.

Ask it from ChatGPT

tasteck connects to ChatGPT over MCP: ask your numbers as a question and the answer comes back in the chat — how many bookings that night, whether settlement has an open discrepancy, how it compares with last week. → Setup guide

Multi-language is built in, the operating surface itself, with your language set during onboarding.

From $34 a month for up to two locations. Thirty days free on every plan, cancel any time. → Pricing

Start this week

1. Print the three sheets and put them where the work is. You cannot print later.
2. Name who declares entry finished. By role, not by whoever.
3. Decide where a discrepancy gets recorded. Not absorbed.
4. Write "stopped at" and "noticed at" separately. The gap is your missing record.
5. Check you can stop a notification flush before you need to.

Read next

• The Incident Log You Do Not Keep
• Cash Room Reconciliation and the Float
• Security and Data
• What a Night Venue POS Cannot Tell You
• Nightclub Software

---

On benchmarks. No uptime figures, recovery-time targets, or outage frequencies appear in this guide. We do not run external uptime monitoring, so we have nothing we could stand behind. A number we cannot evidence is worse than no number.