Three machines,
one verdict

The short answer

A protein-shake machine in the right gym is a good business at roughly half the capital of the alternatives, and it is the only one of the three that can be repeated cheaply. The screen-protector robot has the best payback but needs a quarter of a million rand per door. The refill store is the biggest idea and the smallest return.

So the ranking is not a ranking of machines. It is a ranking of how cheaply you can test the volume assumption. A protein machine costs R95 432 to find out. A screen-protector robot costs R227 310 to find out. A refill outlet costs R665 000 to find out, and it will not tell you anything for a month because it depends on foot traffic through a township or CBD site you do not control.

Run the cheap test first. That is the whole strategy.

A small market growing slower than inflation

South African vending turned over about $156.8m in 2024 and is forecast at $210.9m by 2033 — a compound growth rate of 3.01%. With the rand losing purchasing power faster than that, the real market is flat. Nobody gets rich riding this tide; you get rich taking sites off somebody else.

Three demand pools, only one of which is growing properly

How big could a protein network get?

Assume ~1 000 gyms with enough members to justify a machine. The table shows the annual cash-profit pool at three penetration levels, at our base case of R5 686 net per machine per month.

What each one actually costs to open

All three are imported. Vending machines enter South Africa under HS 8476.29 at zero customs duty, with 15% VAT charged on the added tax value — FOB plus a 10% uplift. The VAT is recoverable if the operating company is VAT-registered; it is working capital, not cost. What is not free is everything around the machine.

One machine, three volumes

Same machine, same price, three different rooms. The cost structure barely moves — only the volume does. That is why site selection is the entire job and the machine is a commodity.

Protein shake dispenser — R38 a shake, 26 trading days

Screen-protector robot — R129 an install, 30 trading days

Refill outlet — R85 basket, 180 customers a day

Price or volume?

Every operator's instinct is to raise the price. The grid says the opposite: this is a volume business, and the fastest way to lose it is to price yourself past a ready-to-drink shake on a shelf.

So: spend the effort on winning the right rooms, and hold the price at a level that beats the R30–R40 ready-to-drink bottle without inviting the comparison. R38 is that number. R45 is a decision to be a supplement shop instead.

What 25 machines looks like

The unit economics are the easy part. A route is a small logistics company with a van, a technician, a manager and a phone that rings at 6pm on a Friday. That overhead is what turns a 41% machine margin into a mediocre business.

Valuation, and why scale is the only lever that matters

A vending route is valued on one of two things: what the steel would fetch in a liquidation, or a multiple of the cash it throws off. In South Africa the first number is close to the second until the fleet is big enough that a buyer believes the earnings are institutional rather than personal.

The three valuation methods, and what each says

• Asset value. Machines at 50–65% of landed cost, the van at 55%, stock at 85%. For a 25-machine route that is about R1.35m — the floor, and roughly what a bank would lend against.
• Earnings multiple. Vending routes in the US change hands at 1.5×–3.0× seller's discretionary earnings. In South Africa, with a thinner buyer pool, expect the bottom half of that band for anything under 40 machines: R1 710 674–R2 138 342 at 25 machines, and the top of the band only once there are contracts, telemetry history and a manager who is not you.
• Discounted cash flow. At an 18% hurdle — reasonable for an unsecured South African venture — the 25-machine route is worth R-230 389 over five years. Negative. The DCF is telling you the truth the multiple hides: on this scale you are buying a job with a resale value, not an asset that compounds.

How this fails, and what to do about it

Six ways this loses money, in the order they are likely to happen.

Staged by dependency, not by calendar

Each stage exists only to answer the question the previous stage could not. Nothing proceeds until its kill criterion is met. Within a stage, everything runs in parallel.

1. Stage 1 — buy the answer

   One machine, one premium gym, ninety days of telemetry

   The only question worth R95 432 is whether a real room does fourteen shakes a day. Pick a Virgin Active Collection, an F45, a high-end independent or a corporate gym in Sandton, Rosebank or Umhlanga. Negotiate 15% commission with no exclusivity so you can leave.

   • In parallel: register the company for VAT, open the Nayax account, set up the import and NRCS Letter of Authority, quote three sites so you know your pipeline before you know your answer.
   • Kill criterion: under 13 shakes a day by day 60 — stop, sell the machine at 60% and take the R35 000 lesson.
   • Pass: 13+ a day, and a second site willing to sign.
2. Stage 2 — buy the route

   Six machines, one van, one part-time technician

   Six sites is the smallest fleet that justifies a vehicle and a technician, and the largest that one person can run without a manager. This stage answers whether the business survives its own overhead.

   • In parallel: negotiate the second supply contract for powder; build the site-qualification checklist from what worked at site one; start the second import order while the first is still landing.
   • Kill criterion: overhead above 45% of gross profit, or fewer than four of six sites hitting target.
   • Pass: six machines at or near base case — around R117 000 of annual cash profit on R468 000 of machine capital, with a manager-free structure that still works.
3. Stage 3 — build the asset

   To 25, then to 60 — and only then does the multiple matter

   Below 25 machines you are buying a job. Above 60 the overhead dilutes to the point where the business earns 56% on machine capital and is worth more than it cost to build. Do not raise money for stage three until stage two is delivering its own cash.

   • In parallel: hire the route manager before machine 15, not after 25; standardise the machine spec so spares are interchangeable; put every site on a written three-year placement agreement.
   • The exit question: a 60-machine fleet at R6 541 731 is a sellable business. Decide now whether you are building cash flow or building a sale — they produce different decisions about contracts, branding and how much you pay yourself.