Andy Wu Ngo
Writing No. 04

Essay No. 04

Surviving became the baseline

August 2026
7 min read
Opinion

There was a kind of night we all recognised and none of us ever wrote down properly.

The weather turns, or a run of bookings lands late, or it’s simply walk-ins, and the number coming through the door has no relationship to the number you rostered for. You work that out somewhere in the first hour. After that it’s the floor: staff working at the top of their ability, a manager moving people into position and briefing them, word going through to the kitchen about when the onslaught is coming and how many covers are behind it. You give the talk. You rally them.

And you get through it. That’s the part that matters. Nearly every time, you get through it.

The week after, there’d be a meeting where we went through that night. Nobody in that room ever asked why the prediction had been wrong.

What got said instead was simpler, and it arrived sounding almost like a compliment. You managed it with that many people. So that many people must be enough.

I want to explain how a room of experienced people reaches a conclusion that wrong, because nobody in it was stupid and nobody was being cruel. I was in that room. Some weeks I was the one saying it.


The room changed over the years. Sometimes five of us, sometimes more. The titles moved around: owner, general manager, head chef, assistant venue manager, whoever was carrying the area that quarter. Different venues, different people, different decades of experience sitting around the table.

The situation never changed.

We’d go through the week that had just finished. Trade, labour, wastage, whatever had gone wrong and whatever had gone unusually right. I was accountable for the labour cost. When the percentage didn’t land, I was the one who explained why.

And I could always explain why.

That was the problem, though it took me a while to recognise it as one. I’m reasonable in a room. I could tell you how trade ran, where the roster was heavy, which section got caught out on the Thursday, what we did to react. It sounded like analysis. Most of the time it was recollection with confidence applied to it.

If I was lucky, I got asked while it was still fresh. The following week, with a set of shift reports to lean on. Those reports were written at the end of long nights by the people who’d worked them, and they were opinionated. Of course they were. That was the record. Somebody’s account of their own shift.


What everyone actually wanted, though nobody put it in these words, was a crystal ball. Tell me how much labour I need on Saturday. Tell me whether next Tuesday is going to happen or not. That thing doesn’t exist, so we did what everyone does, which is hire people with enough years on them to guess well.

Here’s the part it took me a long time to be properly annoyed about.

We already had most of it. Predicted revenue. Expected labour usage. Hours, covers, timings. None of it was exotic and none of it was missing.

It was just held somewhere above me.

Numbers came down from higher up on their own schedule, in a pack, after the period closed. By the time I could see how I’d actually performed, the week I was being asked about was three weeks old and two rosters behind. I was accountable in real time for something I could only see in hindsight.

So you run on judgment. Judgment is real. The best managers I worked with had years of it and were right more often than a spreadsheet would have been. But judgment doesn’t transfer. You can’t hand it to a duty manager on a Tuesday night and expect them to arrive at the call you’d have made.

The people who had the job before me gave vague answers and patched things. I inherited some of those patches. For a long time I read that as them not caring enough to look properly.

I don’t think that any more. The data was available to them too. But to find it, collate it, cross-reference it and actually digest it would have cost more hours than the job left spare. They weren’t being lazy. They were being triaged by their own week.


What Sushi changed was not the data. It was the lag.

Everything I fed it I already had access to, mined out of the same sources that had been sitting there the whole time. The difference was that a job which used to eat an afternoon now landed before the meeting started, and it was more accurate than the version I’d have assembled by hand at the end of a long week.

That’s a smaller thing than it sounds and a much bigger one. The time didn’t go back into my day. It went into thinking. I stopped spending meetings establishing what had happened and started spending them working out what to do about it. Acknowledging a problem and solving one are different activities, and until then I’d mostly been doing the first and calling it the second.

The person it changed most wasn’t me. It was the head chef.

We hadn’t worked well together. When I raised something (a labour number, a wastage figure, an inconsistency in his section) he heard an accusation. He’d get defensive, and I understand why. It was my read of his kitchen, delivered by me, in a meeting where he was outnumbered. Even when I was right, it was still my word about his room.

Then the report started arriving before either of us did.

It wasn’t my read any more. It was just what happened. He could look at the same line I was looking at, at the same time, and neither of us owned it. That sounds like an administrative detail. It rearranged the entire relationship.

There was pushback first. Of course there was. He questioned the numbers, which was fair. I’d have questioned them too, and once or twice he was right and the data was wrong in a way that taught me something about how I’d built it. But the metric didn’t get tired. It didn’t have a bad night. It didn’t have a stake in whose fault Thursday was.

What changed after that was the direction of the conversation. We stopped establishing who was responsible and started working on what to change. Blame is what a room does when it can’t agree on what happened. Once we could agree on what happened, there was nothing left to do but fix it.

An empty function room after a meeting, chairs pushed back at odd angles and a half-wiped whiteboard

Blame is what a room does when it can’t agree on what happened.


It cut both ways, which I hadn’t expected. I’d assumed clear numbers would mostly be a stick. Mostly they turned out to be an argument for spending. Once the changes were tangible and tracked, I could show that a particular Thursday was worth staffing properly, that a window was opening in a trade pattern, that a cost was worth carrying because of what it returned. You can’t ask for money with a feeling.

Which brings me back to those nights, and to something I’d get wrong if I told it simply.

It isn’t that the survival was invisible. The variance was usually right there: the gap between what we’d predicted and what actually walked in told the story clearly enough to anyone who looked at it.

It just wasn’t a line anybody lingered on. It got passed over quickly, often with a smirk. Looked like a busy day. Then on to the next thing.

Because the pressure only ever ran in one direction. Adding staff was treated as something that would breed laziness, a soft decision that needed arguing for. Taking labour out was cost saving, and cost saving never had to justify itself to anyone. One of those was a position you had to defend. The other was just good management.

The attention was lopsided in the same way. A run of good reviews would go by without comment. One bad one would land and there’d be a conversation about standards, and somebody would need to account for what had gone wrong on the floor that night, usually a night that had been thinly staffed for weeks before anyone wrote anything down.

So I’d put it differently now than I would have at the time. The problem was never only that the numbers arrived late. It’s that a number nobody wants to look at is easy to walk past, and a story is much easier to walk past than a number.

What the report changed wasn’t whether the variance existed. It changed whether skipping it was a choice someone had to make in front of other people.


That’s how the ratchet starts, and it only turns one way. The chase to lower labour gets tighter. Nobody studies the mechanics of the night (what actually happened, what it took out of the people who absorbed it) because the result was fine. The result is always fine. And once a venue is in survival mode it is very hard to get it out, because every successful survival is more evidence that the settings are correct.

The expectations grow. The reward for doing the work fast is more work.

None of that lands on a labour report as a problem. The labour percentage looks excellent. It lands somewhere else, months later, in the people who left, in a kitchen that stopped being able to hold quality, in repeat business you don’t notice not returning, in the one review that finally says it out loud.

In this industry that’s a slow death you don’t see coming.

Every successful survival is more evidence that the settings are correct.


I think about those meetings more than I expected to. Not because they were unpleasant, mostly they weren’t, but because of how much experience was in the room and how little of it we spent on the thing sitting in front of us.

Everyone there had watched the same week. Everyone had an account of it.

What nobody had was a version we all had to answer to.

← Back to Andy Wu Ngo No. 04 of 05