Why small swaps cost more

Every row below was measured in the same pass against the same mid-market price, so the only thing changing between them is how much is being sent:

Choosing badly on 25 USDT costs about a fifth of the transfer. Choosing badly on 5,000 costs under two percent. The advice to shop around is worth roughly twelve times more on the small transfer, which is the opposite of how people usually spend their attention.

At 25 USDT only four of eight providers returned a price. At 60 it was six. From 150 upward, seven. Each provider sets its own floor and they are nowhere near each other — on this pair the minimums ranged from about 12 USDT to 120 USDT, a tenfold difference between services offering the same thing.

This is why a small transfer lands badly even when somebody does compare. The comparison is drawn from whoever happens to accept the amount, and the cheapest service may simply not be in the list.

SideShift paid the most of anyone on 25 USDT. On 400, 1,000 and 5,000 it paid the least of anyone. It did not get worse — it charged the same 2.50% under mid-market at every size tested. Everyone else got better as the amount grew, and passed it.

A recommendation of the form “use this provider, it is the cheapest” is therefore incomplete unless it says at what size. Most such recommendations were written about a large transfer, and most people making one are not.

Price the same provider at two sizes and its pricing model becomes visible:

Neither is dishonest and neither is advertised. A fixed cost has to be recovered from somewhere, and on a 25 USDT transfer there is very little to recover it from — which is the whole reason small swaps go badly. The rate you were shown was accurate. It was simply quoted after a charge that does not shrink with the transfer.

The practical consequence: on anything small, a provider charging a flat percentage will usually beat one charging a fixed amount, regardless of which advertises the better rate.

Run the same sweep the other way — Bitcoin into Tether on Tron, at 0.005, 0.02, 0.05 and 0.25 BTC — and the effect disappears. The spread was 1.58% at every size, the best provider 0.80% under mid-market and the worst 2.38% under, identical to two decimal places across a fiftyfold range.

Worth stating plainly rather than explaining away: figures that repeat that exactly are consistent with pricing quoted purely as a rate, with any fixed cost applied later or not reflected in the estimate. The smallest of those four transfers was still about $422, so this sweep does not show what happens to this direction at genuinely small sizes. It shows that between $422 and $21,100, size did not matter.

If you are sending a small amount, compare — the range is enormous and it is the one case where the effort clearly pays. Prefer a percentage charge over a fixed one. And consider whether the transfer needs to be small at all: two transfers of 25 USDT are priced far worse than one of 50.

If you are sending a large amount, the providers converge to within a couple of percent and the choice matters much less than the internet suggests. What matters more at that size is who ends up holding a record of the whole thing.

The pair pages here run this comparison continuously and show every provider's quote beside the mid-market price. The earlier piece on what a swap actually costs covers the same measurement at a single size, including what splitting a transfer adds.

All eight providers were asked for the same pair at each amount in a single pass, so the quotes are comparable to one another. Mid-market is taken from this site's own benchmark, the median of four large exchanges' public ticker endpoints, read in the same session as the quotes.

FixedFloat returned an authorisation error throughout and is absent from every count above, which is why seven providers quote rather than eight. Percentages are computed from the raw payout figures each provider returned, not from any rate they displayed.

Two definitions, stated because they are not interchangeable and the difference is large at the small end. The gap between best and worst is the difference as a share of the best payout, which is what the pair pages on this site report. Distance from mid-market is the shortfall against the benchmark as a share of the benchmark. The first figures in this piece were published on the other convention — as a share of the worst — which read 26.44% where this table now reads 20.91%, and they were brought into line with the rest of the site rather than left to disagree with it.