Two Agents, One Catalogue: Comparing the Fee Stack Line by Line

Comparison basis first: what can be compared and what cannot

A comparison is only as good as the line it is drawn on. The catalogue row is the one line where the two services are genuinely identical: both read the same sellers, so the item price, the option string and the item identity are common ground rather than a point of difference. The conversion implied by the snapshot, roughly ¥6.232 per dollar, belongs to the same common ground, since a service that converts at a different rate changes the item line without changing the catalogue.

Everything above the item line is service-specific and, for this cycle, largely unverified. Fee percentages, the base they are applied to, per-order minimums, payment surcharges, storage windows and intake photograph rules are published in different places, in different levels of detail, and sometimes only inside an account interface. The working assumption for this note is therefore narrow: a figure counts as evidence only when the service states it in public, and an unstated figure is recorded as not verified rather than estimated.

A second rule keeps the table honest. Where a parameter is unknown for one side, the column is marked incomplete rather than scored in favour of whichever service publishes more. Players who answer questions the other side has not been asked should not win by default, and a comparison table that silently rewards disclosure is measuring disclosure.

The six-column table, read left to right

The table below holds the whole worksheet in one view. Each row is one column of the comparison, the third column names the mechanism by which two services can differ even when their headline rates agree, and the last column records what this site could check this cycle.

Reading column one: the fee line, and where a percentage stops meaning anything

Fee structures look like a single number and behave like three. A percentage of item value is the visible part; a per-order or per-item minimum is the part that decides the cheaper half of the catalogue; and a payment surcharge is the part that applies to the total rather than to any item. Two services can quote the same headline percentage and still bill a $300.00 cart $6.00 apart — 3% of $300.00 is $9.00, 5% is $15.00 — which is a small gap against a single order and a large one across a year of orders.

Minimums invert the ranking on cheap goods. At a 3% rate with a $1.00 minimum, the crossing point between minimum and percentage is $33.33; 44 rows of the snapshot sit below $20.00 and 56 more sit in the $20–$34.99 band, so between 44 and 100 of the 218 rows fall at or below that crossing point, where the minimum rather than the percentage sets the fee. Those are illustrative inputs rather than published terms, and they are chosen because they are easy to substitute with the real ones.

Currency conversion works the same way on a smaller scale. At the snapshot’s implied ¥6.232 per dollar, a ¥100 item costs $16.05; at a counter-rate of ¥6.00 it costs $16.67. The gap is $0.62 on one item, about 3.9%, and it is invisible unless the conversion rate is written next to the fee rather than buried in a total.

Reading column two: transit, and why the clock is not only about patience

Transit is the column that appears to be about waiting and is really about storage. Every day a parcel spends in a warehouse consumes the free window that the service grants, and once that window closes the cost of the delay starts appearing on the invoice. A service with a slower handling step can therefore be more expensive than one with a higher fee, without either number being wrong.

Interval evidence for specific lines and destinations was not verified this cycle, so the column is more usefully filled with dated observations of your own than with a borrowed comparative figure. What can be compared without any rate at all is the structure: how many days are free before storage is charged, whether the free window restarts after a merge, and whether a failed delivery consumes it. Those are policy questions with yes-or-no answers, and they belong in the table even when the numbers are missing.

Reading columns three and four: billed weight and the declaration

Billed weight rewards attention because it is decided by a divisor that is easy to state and easy to overlook. For a merged parcel of 40,000 cm³ billed volumetrically, a divisor of 5000 g/L yields 8.00 kg and a divisor of 6000 g/L yields 6.67 kg — a difference of about 20% on the same physical box. Bulky goods carry that exposure most of the time, and the jacket lane’s 23 rows with a median of $47.72, or the accessories lane’s 22 rows with a median of $50.47 and a spread from $3.62 to $190.88, are where a divisor difference is most likely to be felt.

The declaration column is the one buyers treat as a formality and services treat as a liability. The two mechanisms worth comparing are procedural: who enters the values, and whether they can be corrected after the parcel reaches the warehouse. A service that lets a buyer revise a declaration before dispatch is offering something a service that finalises it at intake is not, and the consequences of an inaccurate line fall on the consignee in either case.

Reading columns five and six: returns and support

Return policy is asymmetric in a way that favours the service which states it clearly. The cost of a return leg is usually borne by the buyer, so the value of a good policy is not that it makes returns free but that it makes them predictable: what evidence is required, how long the request window stays open, and whether an unmerged parcel can be sent back without unpacking a consolidated box. An unstated policy is a risk rather than a saving, and neither service’s public return terms were verified here.

Support is measured in a currency that is not time. The useful question is whether the reply can be quoted later, because a written answer that names an outcome converts a dispute into a fact. A fast answer that resolves nothing and leaves no record is worth less than a slower one that can be cited a month afterwards. Since the comparison of the two services rests on this kind of documentation, the support column feeds directly back into the first three.

When the conclusion reverses

Every column has a case in which it stops mattering. Fee percentages lose their force on carts built from the cheap bands, where minimums set the bill; billed weight stops mattering for dense goods that never approach their volumetric weight; declaration rules stop mattering when the value is low and the description is simple; and support quality stops mattering to a buyer who never has to raise a question. A table read without those conditions produces a confident answer to the wrong question.

Three conditions reverse the order most often. A cart of bulky, low-density goods puts the divisor first and the fee percentage third. A cart of small items under $35.00 puts minimum charges first, because that is where the fee is actually set. A high-value cart moves declaration and cover to the front, since an error on those lines costs more than any difference in handling speed.

One reversal cannot be resolved inside the table. Where a parameter is not verified for one side, the honest entry is an unfinished column rather than a default value, which means some comparisons cannot be completed in this cycle at all. The provider profiles are maintained on that basis, with unknown cells labelled rather than inferred, and the ledger carries the fee and transit figures this site has actually checked, each with the date it was checked. A comparison built on those two sources will have gaps; a comparison without them would have errors instead.