Solutions

Meal delivery software for small kitchens

A small kitchen’s advantage is that one person holds the whole week in their head. That works remarkably well, right up until it does not — and the failure is quiet rather than dramatic.

Being honest about when you need this

If you are cooking forty meals a week for a dozen customers you know by name, a spreadsheet and a phone are a reasonable system, and buying software will not make you money. We would rather say that plainly than sell you something early.

What changes is not headcount, it is the number of changes per week. One person can propagate a pause across five documents. They cannot do it eleven times on a Wednesday while cooking, and the ones they miss are invisible until a box goes to a paused customer.

The practical threshold is somewhere between two and four hundred meals a week for most operators, and considerably earlier if your customers pause and swap often. The signal to watch is not revenue — it is how often something has to be corrected after the fact.

Signals you have passed the threshold

  • You have produced meals for someone who paused

    The single clearest sign that change propagation has stopped being reliable.

  • Purchasing and the cook list have disagreed

    Two documents that should derive from one set of selections have drifted, and somebody had to reconcile them.

  • You cannot answer a billing question without opening three files

    If reconstructing what a customer was charged takes an investigation, the record is not a record.

  • Someone other than you now touches the week

    The moment a second person makes changes, the version held in one head is no longer the truth.

  • You have stopped taking holidays

    Not sentimentality — a system only one person can operate is a business with a single point of failure.

What a small kitchen should expect from the software

That it does the propagation and stays out of the way otherwise. A pause entered once should remove the portions from the cook list, adjust the ingredient requirement, drop the labels and remove the delivery stop — in one action, without a checklist.

That it does not require a data-entry role to exist. If keeping the system accurate becomes a job, a small kitchen will simply stop doing it, and an inaccurate system is worse than a spreadsheet because it looks authoritative.

And that setup is measured in days rather than months. A small operator cannot run two systems in parallel for a quarter, so the migration has to be short enough to do over a quiet week.

What it will not do

It will not make you a menu, price your plans or find you customers. Those remain judgement calls, and software that claims to make them for you is selling a fantasy about how this business works.

It also will not rescue bad data. If your customer records have addresses that never found a door and allergen notes typed into a comments field, importing them produces a tidier version of the same problem. Cleaning that up is real work and it happens before a cutover, not during.

Common questions

At what point is software worth it for a small kitchen?
Usually between two and four hundred meals a week, earlier if customers pause and swap often. The honest signal is how often something must be corrected after the fact, not revenue.
Can one person run it?
That is the design assumption. If keeping the system accurate requires a dedicated data-entry role, a small kitchen will stop doing it — and an inaccurate system that looks authoritative is worse than a spreadsheet.
How long does setup take?
Days rather than months. A small operator cannot run two systems in parallel for a quarter, so the migration has to fit into a quiet week.
Will it fix messy customer data?
No. Addresses that never found a door and allergens typed into a notes field import as exactly that. Cleaning them is real work that belongs before a cutover, not during one.

See it at a small kitchen’s scale

The demonstration kitchen runs 27 subscriptions and 63 meals a day — close to the size where propagation starts to fail.