Business

Cloud Kitchen POS in Kenya: Running Delivery-Only (2026)

K By Kev 22 November 2026 10 min read
Share
Business guide

A cloud kitchen POS has to do something a restaurant POS is rarely asked to: operate with no counter, no cashier and no walk-in trade at all. Every shilling arrives through a marketplace, often under two or three brand names cooked by the same line, which changes what the system is for. It is not ringing up sales. It is the only place the business exists as a whole.

On this page▾
  1. Why delivery-only is a harder reporting problem, not an easier one
  2. Setting up a delivery-only kitchen properly
  3. Cloud kitchen mistakes
  4. A worked example: three brands, one line, one loss
  5. How Veira suits a delivery-only kitchen
  6. Frequently asked questions

Why delivery-only is a harder reporting problem, not an easier one

A conventional restaurant has a till that sees most of the business and a delivery tablet that sees the rest. A delivery-only kitchen inverts that: there is no till, so 100% of revenue originates outside the business, in systems the business does not own, under payout terms it did not set.

Multiple virtual brands make it harder again. Two or three listings, each with its own menu and its own marketplace presence, sharing one kitchen, one fridge and one team. The revenue question is per brand and per platform at once — a two-dimensional question — and neither a marketplace portal nor a spreadsheet of deposits can answer it.

The consequence is that cloud kitchens are unusually prone to running a loss-making brand for months. Each brand looks plausible in its own platform portal, because each portal shows revenue and no costs. The food cost and the labour are shared, so the only way to attribute them is to have one system holding every sale, tagged by brand and channel, drawing from one stock position.

Want to see it running in your own business first?

Sign Up

Setting up a delivery-only kitchen properly

The order below front-loads the parts that are painful to change later — brand tagging and recipes especially.

  1. 1

    Record every platform order as a sale, tagged by platform and brand

    Both tags, from day one. Platform alone cannot tell you which brand is working; brand alone cannot tell you which platform is expensive. Retro-fitting brand tags onto months of orders is not realistic.

  2. 2

    Build recipes for every item on every brand

    Shared ingredients across brands is the whole economic point of a cloud kitchen and the reason attribution is hard. Recipes are what let one stock position be apportioned to the brand that consumed it.

  3. 3

    Run one stock position for the kitchen, not one per brand

    There is one fridge. Modelling three is a fiction that will not reconcile. One position, with consumption attributed through recipes, matches reality.

  4. 4

    Put every ticket on one kitchen display

    With no counter, the display is the kitchen's entire interface. Tickets from every platform and every brand in arrival order, marked off as they go, with the brand visible for packing.

  5. 5

    Reconcile each platform payout weekly

    With no cash and no counter, platform payouts are the entire cash flow. A delivery-only kitchen that reconciles monthly is running blind on 100% of its revenue rather than 40%.

  6. 6

    Review brand-by-platform contribution monthly, and be willing to close one

    This is the decision the whole setup exists to support. Gross revenue per brand per platform, less that platform's all-in deduction rate, less the food that brand consumed. Some brand-platform combinations will not clear, and the discipline is to act on it.

Cloud kitchen mistakes

Judging a brand on platform-portal revenue

A marketplace portal shows revenue and no shared costs. Every brand looks viable there. The portal is not lying; it simply has no view of your kitchen.

Skipping brand tagging at launch

The most expensive shortcut available. Without it, brand-level profitability is unrecoverable, and the only fix is starting the measurement from today.

Modelling separate stock per virtual brand

One kitchen has one inventory. Separate positions per brand will never reconcile against a physical count and will generate variance that looks like theft.

Assuming a POS can run the brands on the marketplaces

Each brand's listing, menu and pricing live in each marketplace. Veira records the resulting sales and holds the stock behind them; it does not publish or update marketplace menus for any brand.

Treating prep time as a kitchen-wide average

With several brands sharing a line, one slow item on one brand can drag every ticket. The average hides it; per-item and per-brand times find it.

A worked example: three brands, one line, one loss

Worked example

A delivery-only kitchen in Nairobi ran three virtual brands — a grill concept, a wings concept and a rice-bowl concept — across two marketplaces, from one line of four people. Combined revenue was healthy and growing, and the operation was losing money in a way the owner could not locate.

All three brands looked fine in the platform portals. The problem only became visible once every order was recorded as a sale tagged to both brand and platform, with consumption attributed to one stock position through recipes. The wings concept was the volume leader by order count and the loss. Its food cost was the highest of the three, its average order value was the lowest, and it was concentrated on the platform with the higher all-in deduction rate. Those three facts were each individually invisible and jointly decisive.

The kitchen display data added a second finding. Wings tickets took the longest to prepare and were arriving in clusters at peak, which was pushing the other two brands' tickets late. So the loss-making brand was also degrading the profitable ones. The wings brand was retired on the expensive platform and repriced on the other, and the line got its peak back. None of that was available from revenue figures.

Business impact

Trading without eTIMS-compliant tax invoices risks KRA penalties, blocked VAT input claims for your customers, and receipts a business buyer cannot expense.

Veira signs every sale to KRA eTIMS automatically, so each receipt is compliant the moment it prints, with no separate device to reconcile.

How Veira suits a delivery-only kitchen

Veira integrates with Uber Eats, Glovo and Bolt Food, recording each order as a sale tagged to its channel, drawing from one stock position, reporting by channel, and showing the ticket on the kitchen display. For a kitchen with no counter, that chain is the business: it is the only place where every brand and every platform is visible at once.

Veira runs without a physical cashier and does not need a counter — it works on a Veira terminal or a phone the kitchen already has, and keeps working offline, syncing to KRA when connectivity returns. Ingredient consumption is tracked behind sellable items, so one stock position can be set against what each brand sold. Prep times come out of the kitchen display, per item and per channel. Every sale issues a compliant KRA eTIMS invoice, which matters more here than anywhere: with no cash trade, platform statements would otherwise be your only record of turnover.

The limits are the same as elsewhere and worth repeating for this case, because a cloud kitchen depends on marketplaces entirely. Veira does not create or manage brand listings, menus or pricing on any marketplace. It does not accept orders on a platform's behalf, does not dispatch riders, and does not reconcile payouts — it gives you the gross sales record to reconcile them against.

Frequently asked questions

What POS does a cloud kitchen actually need?
One that records every marketplace order as a sale tagged by platform and brand, holds one stock position for the kitchen, shows tickets on a kitchen display, and runs without a counter or cashier. Payment handling matters far less here than reporting, because the platforms collect the money.
Can one POS handle several virtual brands from one kitchen?
Yes, and brand tagging is the critical part. Tag every sale with both brand and platform from launch — it is the only way to answer which brand works, and it cannot be reconstructed afterwards.
Should each virtual brand have its own inventory?
No. One kitchen has one fridge, so one stock position. Attribute consumption to brands through recipes instead. Separate positions per brand will never reconcile against a physical count.
How do I tell which brand is profitable?
Gross revenue per brand per platform, less that platform's all-in deduction rate, less the food that brand consumed. All three parts are needed — the shared costs are exactly what the platform portals cannot see.
Can a cloud kitchen use a POS with no cashier?
Yes. Veira runs on a terminal or a phone and does not require a counter or a cashier role. For a delivery-only kitchen the POS is a record and a kitchen interface rather than a till.
Do delivery-only sales still need eTIMS invoices?
A sale is a sale. Recorded in Veira it issues a compliant KRA eTIMS invoice automatically, which is particularly important with no cash trade — otherwise platform statements are your only evidence of turnover.
What should a cloud kitchen operator review daily?
Orders and gross revenue by brand and platform, prep times at peak, and stock on your fastest-moving ingredients. The brand-level contribution review is a monthly job, not a daily one.

A delivery-only kitchen lives entirely inside other people's systems unless it keeps its own record. If you are running more than one brand off one line, book a demo — the brand-by-platform contribution number is usually the one that changes a decision.

Need help, or want to talk it through first? Chat with us on WhatsApp
Terms explained

Keep reading

See all Business guides

Veira for your business

Want the full picture for this trade? Veira for restaurants covers the counter, the stock and the KRA side together.

On the Veira side of this: how orders reach the kitchen without a printed chit.

Browse Veira by business type · Who we serve