You are renting footfall

You have seen them everywhere: the coin-operated massage chair, in mall corridors, petrol stations, hospital waiting areas, transport terminals, outside convenience stores. ฿10 or ฿20 for a few minutes. Coins or, increasingly, a QR code.

And somebody is selling you one.

Thai vendor material, checked again in September 2026, puts a machine at around ฿85,000 — some at ฿67,900 against a list price of ฿89,900 — with a projection of ฿400 a day, ฿12,000 a month, ฿144,000 a year, against electricity of roughly ฿2,100 a year.

Payback in about seven months.

Every one of those figures comes from a person who wants you to buy a machine, and I will keep saying that, because it is the most important fact about most of the available information on this business.

What ฿400 a day actually requires

Take the projection seriously for a moment, because it is easy to nod at.

At ฿20 a session, ฿400 a day is twenty paying users, every day. Not twenty on a Saturday. Twenty on a wet Tuesday in September, twenty on the day the machine is dusty, twenty every day for a year.

That is one paying customer roughly every forty minutes across a twelve-hour trading day, sustained, with no seasonality, no downtime, no vandalism, no maintenance day, and no month where the site is quiet.

At ฿10 a session it is forty. The same vendor material advertises both prices and quotes the same daily figure.

And the arithmetic is rounder than it looks. ฿400 a day becomes ฿12,000 a month, which is a thirty-day month, and twelve of those make ฿144,000 — five days short of a year. It is not a forecast that has been built. It is a number that has been multiplied.

The same four figures recur across sellers, unchanged. ฿85,000, ฿400, ฿12,000, ฿2,100. That is not several traders arriving independently at similar estimates. It is one set of numbers being passed along, which is worth knowing before you treat their agreement as corroboration.

It is not impossible. It is the top of the range presented as the middle of it.

The number that is not in the model

Look at what the vendor arithmetic contains: machine cost, revenue, electricity.

Now look at what it does not contain.

What the site owner charges you.

A massage chair is not in your house. It is in a mall corridor, a forecourt, a hospital lobby — somebody else’s floor, and that somebody has an opinion about what a square metre of their footfall is worth.

That arrangement is the entire business, and it is the one input the seller cannot quote you, because it is not theirs to quote.

An earlier version of this article went on to say that the number is therefore unpublished — no range, no typical structure, nothing. That was wrong.

It is published. Just not by anyone selling you a machine

Thai institutions let their floor space by public notice, and the notice states the price.

On 30 July 2024 the Faculty of Medicine Ramathibodi Hospital, Mahidol University, published an invitation to bid for the right to install one vending machine in its Rama Pavilion building in Phayathai. The notice carries a rate table. Rent: not less than ฿2,000 a month. Term: three years, running from 1 November 2024 to 31 October 2027. A letting fee of twice the monthly rent for each year of the term. A deposit of three times the monthly rent. Electricity at ฿5 a unit, metered. Water at ฿20 a unit. Other charges as the Faculty decides.

That is a vending machine, not a massage chair. But three weeks earlier, on 12 June 2024, the welfare committee of the Golden Jubilee Medical Centre — part of the Faculty of Medicine Siriraj Hospital, also Mahidol — published an invitation for one operator to install coin-operated massage chairs in the corridor outside ward 4B on the fourth floor.

Its terms are the same instrument. Three-year lease. A letting fee of twice the monthly rent a year, and a deposit of three times the monthly rent. Electricity at ฿6.50 a unit, metered, and the centre supplies the space but not the electrical installation — the operator pays to bring power to the spot. The operator carries the property taxes, the value-added tax and the lease registration fee. And if the chair or its coin mechanism stops working, the operator must repair or replace it within twenty-four hours of being told.

The applicant must be at least twenty-five, must have run a shop for at least three years, and must operate it personally.

A third document, from a different institution, prices the floor rather than the machine. On 24 June 2026 Ban Phaeo Hospital, a public organisation in Samut Sakhon, invited operators for three shop units in its eye hospital: 8.09 square metres at not less than ฿13,000 a month, 11.54 square metres at not less than ฿12,000, and 23.40 square metres at not less than ฿18,000. That is between about ฿770 and ฿1,600 per square metre per month, as a floor, on a three-year term.

A chair, on its manufacturer’s published specification, is 148 by 83 centimetres standing and 168 by 83 reclined — and it has to recline to work. Call it one and a half square metres before anyone walks past it. At Ban Phaeo’s published rates that floor would be somewhere around ฿1,100 to ฿2,400 a month, and Ramathibodi’s ฿2,000 floor for a single machine sits inside that range.

Two institutions, two different things being let, landing in the same place. It is weak evidence and it is the only evidence there is. But it is evidence, and it is a great deal better than a blank.

Against a vendor projection of ฿12,000 a month, a floor of ฿2,000 is a sixth of the revenue before a single bid has been made. And a floor is where the bidding starts.

You do not get the site by paying the most for it

This is the part the arithmetic cannot show you, and it is in both notices.

Ramathibodi asks for three sealed envelopes. The first is qualifications. The third is price — and price has two lines: the monthly rent, and any other benefit given to the Faculty that is not a donation — the example the notice itself gives is a share of revenue from sales.

The Golden Jubilee notice asks for three lines too: the monthly rent, benefits for the medical centre’s own staff, and a donation, if any.

So the rent is not the offer. The rent is the first line of the offer.

Then comes the clause that decides everything. Ramathibodi reserves the right not to accept the highest offer, or any offer, or to cancel the process entirely, for its own benefit; the committee’s decision is final; and no bidder may claim damages. The Golden Jubilee committee says the same thing in its own words: it will select whichever applicant it considers appropriate, its decision is final, and no applicant may bring suit or make any claim. Ban Phaeo scores its applicants out of a hundred and gives the money thirty of those points — the same weight as experience, ahead of service quality and the price list — then reserves the right not to take the highest bid anyway.

And Ramathibodi has one clause I did not expect to find. If the winning offer is so high that the Faculty doubts it can be delivered, the bidder must explain themselves and produce evidence that they can actually perform; if the explanation is unconvincing, the Faculty may reject them.

The landlord has written into its own notice the expectation that bidders will offer more than the site can earn.

There is one more. After selection, and before signing, the Faculty reserves the right to negotiate further conditions beyond the ones you offered.

Why the money ends up with the landlord

Here is the structure, and it is the answer to the question in the headline.

The machine is a commodity. Anyone with ฿85,000 can buy one, from several suppliers, this week. There is no scarcity, no barrier and no proprietary advantage.

The site is not a commodity. There is exactly one spot beside that particular escalator, and the person who controls it can only sell it once.

When a commodity input meets a scarce input, the return accrues to the scarce one. That is not a Thai phenomenon or a moral claim; it is what happens in every market where one side can be replaced and the other cannot.

Usually that is an economist’s argument and you are asked to take it on trust. Here it is a procurement document. The scarce side sets the floor, defines the currency the bid is paid in, scores you on criteria of its own choosing, keeps the right to refuse the best offer, keeps the right to renegotiate after you have won, and rules out any claim against it. The commodity side turns up with a machine.

So if a chair in a good site earns well, the site owner learns that — and at renewal, the fee rises to absorb it. If it earns badly, the fee does not fall, because the landlord has other people asking.

You do not own a machine business. You own a machine, and rent a business.

Article 116 The bridge is the asset found the same shape in Bangkok’s malls, where the thing actually worth owning turned out to be the bridge rather than the building. This is that finding at ฿20 a go.

The landlord asks for the information you cannot get

The second Ramathibodi envelope is the technical one, and its contents are the sharpest thing in this article.

The bidder must submit a service plan, a maintenance plan, a restocking plan, a cleaning plan, a plan for the upkeep of the equipment — and a plan for refunding or compensating users who have paid and then had the machine fail or misbehave.

Read that against what a prospective buyer can find out. How long does a machine last in public use. What does a repair cost. How often does one break. What proportion of takings goes back out as refunds. None of it is published, and the landlord requires you to have written it down before it will let you through the door.

The information does exist. It exists inside operators, and the only party systematically collecting it is the party on the other side of the table.

The Thai part, which is genuinely interesting

None of the above is unique to Thailand. This bit is.

Article 50 established that Thailand skipped credit cards and went from cash to QR, and that PromptPay reaches almost everybody. That did something specific to unattended retail.

A coin machine requires a cash operation. Somebody drives a route, opens boxes, counts, banks, reconciles, and absorbs the losses when a box is forced. In most of the world that collection operation is the majority of the cost and all of the risk of running unattended machines, and it is why the business only works at scale.

A QR machine has none of it. The money arrives in an account the moment it is paid. No route, no float, no coin box worth breaking open, no reconciliation.

Thailand’s payment rails did not just change how people pay. They changed which businesses one person can run, and a single-machine operator is one of them. That is why you see so many of these, owned by so many different people.

And the Golden Jubilee notice shows you what that permission costs. The operator must be a natural person who runs it themselves, of at least three years’ standing, on call to fix the machine within a day. The rails let one person do it. The lease requires that one person to be there.

And the same thing that made it easy made it expensive

The QR record cuts both ways, and this is the part nobody selling you a machine will mention.

Article 128 You can see the gearbox found that markets change when the goods can be verified. A QR-paid machine produces a verifiable revenue stream, and a landlord who can see or infer real takings can price the site against them.

In a coin world, the site owner was guessing. You could plausibly claim the chair did modestly. In a QR world your revenue has a shape that is much harder to be vague about, and any surplus above a commodity return is legible to the person who can raise your rent.

The technology that removed the cash-handling cost also removed your information advantage. Both effects are real and they point in opposite directions.

The one cost the model does contain is the one that does not matter

The vendor model includes electricity, and electricity is the cheapest thing in it.

A manufacturer’s published specification for a coin-operated chair gives 110 watts at 220 to 240 volts. Twenty sessions of five to ten minutes is between one and two thirds and three and a third hours of running a day — under four tenths of a unit of electricity.

At the rates those two hospitals charge their own tenants, ฿5 and ฿6.50 a unit, that is somewhere between about ฿0.90 and ฿2.40 a day, or ฿330 to ฿870 a year. The vendor figure of ฿2,100 is two to six times it.

The session length is my assumption, because vendors do not state it, and the specification does not give a standby draw, which could account for some of the gap. The direction is not in doubt. The one cost the model troubles to include is a rounding error, and it is stated generously — which reads as prudence and is not.

Note also who is selling you the electricity. In both notices the landlord sets the unit rate, and one of them sets it higher than the other. You are not buying power from the utility. You are buying it from the person you are already paying rent to.

What I still cannot tell you

I have no independent revenue figures. Every number in circulation for what one of these machines takes is a vendor projection, and they are all the same projection.

I have no site fee for a massage chair in a commercial location. What I have is a floor price for one machine site at one public hospital, a lease structure for chairs at another, and shop rates at a third. Malls, petrol stations and transport terminals do not publish, and they are where most of these machines are.

I do not have the outcomes. Public bodies publish what they will accept; they generally do not publish what they got. The floor is public and the clearing price is not, which means the gap between them — the thing that decides whether this works — is exactly the part still missing.

And the two chair-and-machine documents are one institution. Ramathibodi and the Golden Jubilee Medical Centre are both Mahidol bodies letting space under the same 2018 university regulation, so their agreement tells you about Mahidol’s drafting, not about the market.

I have no failure rate, no machine lifespan, and no maintenance costs — the vendor material offers a ten-year motor warranty and two years on the structure, which is a seller’s claim about a seller’s obligation, not evidence about how long a chair survives in a corridor.

And I have no count of machines in service, so nothing in this article should be read as a statement about the size of the trade.

A market where most of the available information comes from sellers is not a market with good information. Article 80’s finding applies precisely: there is a published price for the machine and almost nothing published about the returns, and the person on the wrong side of that gap is the buyer.

So, is it worth investing in?

I am not a financial adviser and this is not advice. It is a structure for asking better questions, and the documents above answer several of them better than any vendor will.

The site agreement is the asset, so read one before you buy anything. The two hospital notices are public, they are four pages each, and they show you what an institutional landlord actually asks for: a term, a floor, a deposit, a letting fee, a metered utility rate, a repair clock, a maintenance plan, and a reserved right to say no. A private landlord’s terms will be less formal. They will not be more generous.

Notice what the rent is denominated in. If the offer includes a revenue share, a benefit in kind or a donation, the headline rent is not the price.

Notice who bears the machine’s failures. In one of these notices it is twenty-four hours, in writing, at the operator’s cost, plus a refund plan for the users.

And notice the clause that says the highest offer need not win. It is there because the landlord has seen people bid past what the site can earn — a risk the landlord has taken the trouble to write down, and the seller’s projection has not.

The honest summary: this is a location business with an equipment cost attached, and it is being sold as an equipment business with a location detail attached. Everything follows from which of those you think you are buying.

Common misconceptions

“It’s passive income.” It requires a site relationship, maintenance, and cleaning. One published lease requires the operator to manage it personally and to fix a failed machine within twenty-four hours. The passivity is in the marketing.

“The machine is the investment.” The machine is a commodity available to anyone. The site is the scarce input, and returns accrue to scarce inputs.

“฿400 a day is a normal figure.” It is a vendor projection requiring twenty paying users every day with no seasonality or downtime — forty at ฿10 a session — and the same four figures appear across sellers unchanged.

“Nobody publishes the site fee.” Thai institutions publish it in their own procurement notices. One Bangkok teaching hospital set a floor of ฿2,000 a month for a single unattended machine on a three-year term.

“The site goes to whoever pays most.” Both notices reserve the right to refuse the highest offer, and one requires a bidder whose offer looks too high to prove they can deliver it.

“QR just makes it convenient.” It removes the cash-collection operation that makes unattended retail hard, which is why single-machine ownership is viable here. It also makes your revenue legible to the person who sets your rent.