Central Kitchen Software: What Actually Runs a Commissary (Singapore, 2026)

Search "central kitchen Singapore" and you will find pages of industrial units for rent, listings for kitchen space in Woodlands and Tuas, and agents offering to sublet cold room capacity. Almost nothing on what happens inside one once you have signed the lease.

That gap matters, because the central kitchen is where multi-outlet F&B operators in Singapore quietly lose money — not on rent, and not on food cost in the way most people mean it. This article is about the operating discipline, not the property. Software comes in at the end, and only because there is a specific point at which the discipline stops being sustainable by hand.

The failure mode: outlet number three

The pattern is consistent enough to be predictable.

At one outlet, you are in the kitchen. How much to prep, what is left over, what sold out by 7pm — all of it lives in your head, and your head is a perfectly adequate system at this scale.

At two outlets, it moves to a WhatsApp group. "How many kg today", "send me a photo of the chiller". Clumsy, but it works.

At three, it stops working. Not gradually — the coordination cost of an informal system rises faster than the number of outlets, because every outlet has to stay in sync with every other one. The symptom operators report is oddly specific and almost always the same sentence: _every outlet is profitable, but there is no cash in the account._

That sentence is not an accounting mystery. It is what unrecorded loss looks like on a P&L.

The seven daily checkpoints

Here is the operating cycle a commissary serving multiple outlets needs to run, in time order. Read it as a diagnostic — the question is not whether you have software, it is which of these seven you are actually doing.

1. Receiving

Weigh and log on arrival. Not at noon, not when someone gets around to it.

The critical detail: log what you received, not what you ordered. Those are two different numbers, and the gap between them is the entire point of the exercise. Logging the purchase order instead of the goods received deletes the signal at the source — you have created a record that can never tell you a supplier short-delivered.

2. Prep start

Lock the recipe and the batch size before anyone starts cooking.

Three chefs cooking the same dish three ways produces three different costs for the same menu item, and no amount of downstream reporting can un-mix that. Cost consistency is a constraint you apply at the point of work, or not at all.

3. Prep complete — _the first skipped checkpoint_

Record input consumed against output produced.

This is the one nearly everyone skips, and it is the single largest blind spot in most Singapore commissaries. Skip it and your data set is: goods in, sales out, and an unlit box in between. Every yield problem, every over-prep, every quiet shrinkage lives in that box, permanently unattributable.

You will still know your food cost percentage. You will not know _why_ it moved.

4. Sorting by outlet

One pack, one outlet, one sheet — with "should send" and "actually sent" recorded side by side.

The variance between those two columns is worth more than either number on its own. Record only the intended quantity and you have made the same class of error as checkpoint 1. Catch the gap in the kitchen at 2pm, not on the phone at 6pm with an outlet manager who is now short for dinner service.

5. Cold-chain temperature

Take and log a temperature before loading.

This is partly SFA compliance, and worth doing on those grounds alone. But the practical argument is narrower: on the day something goes wrong, having a temperature record and not having one are two completely different conversations — with your customer, your insurer, and the regulator.

6. Dispatch handover

Driver signs. Outlet signs. Both ends.

If stock goes missing between the commissary and the outlet, a two-sided handover tells you which leg it went missing on. A one-sided one tells you only that it is gone.

7. Same-day review

Prepped versus dispatched versus left over — reviewed the same day.

Same rows, same fields, same query as your month-end report. The only variable is latency, and latency is precisely what decides whether the information is still actionable. Reviewed at month end, this is accounting. Reviewed same day, it is control.

Two of these are optional in practice. Two are not.

Checkpoints 3 and 7 are the ones that get dropped first, and they are the two that make the rest cohere.

Without prep complete, you have no consumption data, so checkpoints 1 and 4 produce numbers with nothing to reconcile against. Without same-day review, every other record is written but never read — you have paid the full cost of the discipline and captured none of the benefit.

An operator running all seven with pen and paper is in better shape than one running a system that logs five of them.

When does software actually matter?

Honestly: later than most vendors will tell you.

None of the seven checkpoints requires software. At two outlets, a clipboard and a shared spreadsheet handle this without much strain, and plenty of well-run Singapore F&B businesses operate exactly that way.

The constraint is not capability, it is sustained effort. Past roughly three outlets the recording burden exceeds what anyone will keep up under service pressure — and what happens then is not that the records get worse, it is that checkpoints 3 and 7 get skipped entirely, because they are the two with no immediate consequence for skipping. Nobody chases you at 4pm for a yield log.

That is the actual failure mode. Not the tooling. The sustaining.

Software helps at that point for one structural reason: it can make the recording a by-product of the work rather than an additional task. When marking a production batch complete _is_ the consumption log, checkpoint 3 stops being something to remember. That is the whole mechanism — there is no cleverer version of it.

Get the discipline right first, then automate it. In that order. Systems deployed onto an undefined process reliably produce an expensive, well-documented version of the same mess.

Where MakanCloud fits

For transparency about our own position: MakanCloud is a Singapore restaurant operations platform, and the central kitchen module covers checkpoints 1 through 7 — procurement and goods receipt, production work orders with locked recipes and automatic stock deduction on completion, picking and sorting by outlet with should-send/actually-sent variance, outbound dispatch with temperature checks, and a same-day operations dashboard. It is used daily in a real Singapore restaurant group, which is why the checkpoint list above is descriptive rather than aspirational.

Two things worth stating plainly, since this article argues for honest assessment:

We are not the only system with back-of-house capability. Several competitors offer procurement, inventory, and supplier management modules. The distinction we would draw is narrower and more specific: production work orders, sorting, outbound dispatch, and temperature logging — the stages between "stock arrived" and "stock left for the outlet" — are less commonly modelled, and those are exactly where checkpoints 3 through 5 live. When evaluating any vendor, including us, ask which of the seven stages are covered rather than accepting "supply chain management" as an answer.

And our pricing is published on our site — four tiers, S$59 to S$299 per month — so you can work out your cost before talking to anyone. We mention this not as a claim about being cheaper than alternatives (depending on your outlet count, we may not be) but because several systems in this market only quote after a demo call, and being able to budget before that conversation is worth something.

The short version

• The central kitchen problem in Singapore is an operations problem that gets discussed as a property problem
• Seven daily checkpoints; checkpoints 3 (prep complete) and 7 (same-day review) are the ones that get skipped and the ones that matter most
• "Every outlet is profitable but there is no cash" is the diagnostic sentence — it usually means checkpoint 3 is missing
• Pen and paper is genuinely fine at two outlets. Past three, the skipping starts
• Define the process first; automate second

If you are opening outlet three and want to talk through which checkpoints you are missing, we are happy to have that conversation whether or not it ends with our software.