The number everyone quotes is the wrong one

Ask what a payment provider costs and you will be told a percentage. Five percent, four percent, two percent. It is the number on the pricing page and the number people compare. For most businesses it is also the number that matters, because most businesses sell things for twenty or fifty or two hundred units of currency, and at those prices a percentage is genuinely what you pay.

We wanted to sell a day pass — twenty-four hours of the paid features, one charge, nothing recurring. The natural price was about a dollar. At a dollar, the percentage stops being the number that matters, and almost nobody warns you about this before you have already designed the product around the price.

What a merchant of record actually charges

A merchant of record sells your product as the legal seller, handles sales tax and VAT in every jurisdiction, and pays you the remainder. For a small software business selling internationally this is a genuinely good deal, because the alternative is registering for tax in places you have never been. The pricing is nearly always a percentage plus a fixed amount per transaction.

Two published examples, both checked on the day this was written: Lemon Squeezy charges 5% + 50¢, and Creem charges 3.9% + 40¢. Those are ordinary, competitive rates. Neither is trying to overcharge anybody.

Now put a one-dollar sale through them. Lemon Squeezy takes five cents of percentage and fifty cents of fixed fee: 55 cents on a dollar. Creem takes about four cents and forty cents: 44 cents on a dollar. You have sold something for a dollar and kept roughly half of it, before you have paid for a single byte of bandwidth.

The percentage was never the problem. On a one-dollar sale, Lemon Squeezy's famous five percent is five cents — trivial. The fifty cents is the entire story, and it is the same fifty cents whether you sell for one dollar or fifty.

Which is why the same product is $2, not $1

A fixed fee does not scale, so the only lever you have is to spread it over a larger sale. That is not a pricing strategy anybody enjoys explaining, but the arithmetic is blunt:

At $1, a 50¢ fixed fee is 50% of the sale.
At $2, the same 50¢ is 25%.
At $5, it is 10%.
At $20, it is 2.5% — and now the percentage is finally the number that matters again.

Our global day pass is two dollars. It is not two dollars because we did market research and discovered that two dollars is what the market will bear. It is two dollars because one dollar meant giving away close to half of every sale to move fifty cents, and we would rather charge a fair price once than quietly make the product worse to justify a bad one.

India is a genuinely different market, for a structural reason

In India the same day pass is ₹20 — roughly a quarter of the dollar price. That looks like ordinary purchasing-power pricing, and partly it is. But the reason ₹20 is possible at all is not about purchasing power. It is that UPI, the rails most Indians actually pay with, has had no fixed per-transaction floor.

Since January 2020, Indian law has prohibited a merchant discount rate on UPI and RuPay debit transactions. Section 10A of the Payment and Settlement Systems Act, 2007 made it illegal to charge merchants for accepting them. That is why a ₹20 charge over UPI is viable when a $1 charge over a card network is not: on a card there is a floor of thirty or forty cents that exists regardless of the sale, and on UPI there has been no floor at all.

A payment gateway still charges you for the integration, support, settlement and reconciliation — the UPI rails being free to the merchant does not make a gateway free. But the gateway's fee is a percentage, and a percentage of ₹20 is small. The structure is what changes, not the generosity.

That law changed last month

The Taxation and Other Laws (Amendment) Bill, 2026 was introduced on 4 August 2026 and has since passed Parliament. It amends the same Section 10A that had enforced the blanket zero-MDR rule for six years, replacing the fixed statutory reference with a provision letting the central government specify, by notification, which electronic payment modes may not be charged for. The prohibition is no longer absolute.

What the Bill does not do is set a rate, name a threshold, or define which merchants would pay. Finance Minister Nirmala Sitharaman told the Rajya Sabha that no MDR framework has yet been finalised, and RBI Governor Sanjay Malhotra has described talk of the fee as being at a premature stage. The operating rules — who qualifies, what triggers the charge, whether it is a percentage or a capped amount, how refunds are handled — fall to the UPI and Services Steering Committee chaired by NPCI. None of them has been settled.

The government has said consumers will not be charged at all, that the majority of merchant transactions will stay free, and that any MDR would apply only above a threshold, to a limited set of merchants, at a rate below card MDR.

₹2,000 is not a threshold, and you will see it quoted as one

One number dominates the coverage, and it is worth being precise about, because it is being repeated as though it were settled. ₹2,000 has not been notified as an MDR threshold.

It comes from somewhere else entirely: the government's incentive scheme for low-value BHIM-UPI person-to-merchant payments, under which small-merchant transactions up to ₹2,000 have earned an incentive of 0.15% of transaction value since FY2021-22. As one analyst put it to Business Standard, it was originally a fiscal-targeting line for subsidy, not a payments-cost line. Reuters has reported ₹2,000 as one option under discussion for a transaction threshold applying to larger merchants. That is a long way from a rule.

There is a coherent argument for a threshold somewhere around there, and it is worth understanding because of what it says about intent: only about 4% of person-to-merchant UPI transactions exceeded ₹2,000 in FY2025-26, but that slice carried roughly two-thirds of the total value. A high threshold captures most of the money while leaving almost every actual payment untouched — the vegetables, the tea, the ₹20 purchases. On the evidence available, the design is specifically trying to avoid micro-payments.

What we are doing about it: nothing, for now

Two separate things would both have to go against us. Our prices are ₹20 and ₹99 — 1% and 5% of the figure under discussion — so a transaction-value threshold anywhere near ₹2,000 does not reach us. And the finance minister has said small traders are not the target, which is the second shield. Repricing now would mean reacting to a framework that does not exist.

There is one specific shape this could take that would matter, though, and it is the same one this whole article is about. A percentage MDR is close to harmless at our prices: even a full 1% on ₹20 is twenty paise, against the gateway fee we already pay. A fixed amount per transaction is what kills a micro-price, exactly as fifty cents kills a one-dollar sale. So when the framework is published, the first question worth reading is not the rate — it is whether the charge is a percentage or a flat fee.

We think a flat fee is unlikely, because it would invert the purpose of the policy it replaces. We are writing it down anyway, because our ₹20 price rests on a rule that has just been made amendable, and the honest version of that sentence is not this is fine but this is fine for reasons we can name, and here is the signal that would change our mind.

What we would tell anyone pricing something cheap

Work out the fixed fee as a percentage of your price before you fall in love with the price. Providers advertise the percentage because the percentage flatters them; the fixed fee is what decides whether a micro-price is a business or a donation.

Then check it per rail rather than per provider. The mistake we nearly made was choosing one payment provider for everything, because that is obviously simpler. It also would have meant either charging Indian customers a card-shaped price for a UPI-shaped product, or charging everyone else a price that only makes sense on rails they cannot use. The same provider can be the right answer for one of your price points and clearly the wrong one for another — that is a normal outcome, not a sign you have analysed it badly.

And if the arithmetic says a price cannot work, raise the price or change the product. Do not ship the price anyway and hope volume fixes it. Half of every sale is a lot to give away for the convenience of not having done the multiplication.

Rates quoted here were taken from each provider's own public pricing page on 5 September 2026 and will drift; check them before relying on them. This is a description of how we priced our own product, not tax or financial advice — for anything involving GST registration, exports or an LUT, talk to a chartered accountant.

Why a ₹20 Product Cannot Be Sold for $1