Fee Drift: Six Months of Service Fee Snapshots
1,517 words · about 8 minutes · note 25 of 40 · Step 3
Fee drift is usually discussed as a rate change, which is the one form of it that is easiest to check and least often responsible for a bigger invoice. What moves a total between two seasons is more often the base a percentage is applied to, the moment a per-parcel charge is triggered, or the weight band a parcel falls into. This note treats six months of service fee snapshots as a ledger problem: which rows carry a date, which values are honestly unverified, and what a one-point move would cost on the median listing in this site’s own sample.
What six months of snapshots means on this site
A snapshot is a value with a date attached, and a ledger is a set of such values kept in the order they were taken. The value matters less than the date, because a fee figure without one cannot be compared with anything: there is no way to tell whether it describes the period before or after a change. This site keeps its fee and transit rows in that dated form on the ledger page, and the ledger’s change log records what moved and why.
Honesty about coverage is part of the format. This site can derive price statistics from its own catalogue sample, and it re-checks the editable fee template that the tools use, but it cannot read any service’s current rate card from a public page — that is why every default in the template carries the status not verified and a note telling the reader to replace it with what their own account displays. A six-month series of real rates is therefore not something this note can supply.
What it can supply is the arithmetic of drift: how much a single point of commission costs at different order values, which lines are sensitive to a change of base, and which indicators would show a structural change if one occurred. Every figure below is either taken from the dated sample or shown as arithmetic on a template value, and each is labelled as such.
The six-month record, row by row
Read as a ledger, the six-month record is a table with a status column. One row carries a date this site can stand behind: the snapshot of 218 listings taken on 2026-09-29, together with the re-check of the fee template on the same day. Five neighbouring period rows contain no dated fee observation, and their honest status is that they are unverified placeholders to be replaced with the figures an account actually charged.
That is an uncomfortable table to publish, and it is the correct one. Filling the empty rows with plausible-looking percentages would create a series that no reader could check and that this site could not defend, and a fabricated trend is worse than an admitted gap because it invites decisions. The same discipline applies to the rate a reader pays: it is displayed inside an account, and outside that account it is unverified.
The one populated row still does real work. It fixes the price distribution against which every percentage is calculated, so a reader can see whether a fee change is large or small relative to the amounts it applies to. A one-point commission movement on a $15 listing and on a $150 listing are the same rate and very different money; the distribution is what makes that difference visible.
| Period | What the record holds | Status | What the row can support |
|---|---|---|---|
| Earlier periods | No dated fee observation kept by this site | not verified | Nothing; to be replaced from your own account figures |
| 2026-09-29 | Fee template re-checked; 218-listing price snapshot taken | verified as a dated record of the sample | The price distribution and the template’s fee bases |
| Any single account | The rate card displayed when the fee is charged | not verified outside that account | Only the figures that account displays |
| Next period | Pending; the indicators are listed in the final section | not verified | Nothing until a dated row exists |
Comparing this cycle with the previous one
A like-for-like fee comparison needs two dated rate rows and one unchanged reference order. This site holds one dated row, so the comparison it can make is structural rather than numeric: how the sample’s distribution is shaped now, and which parts of a fee schedule are capable of moving an invoice without any headline rate changing.
The distribution itself is the most useful stable reference. Across the 218 listings, the median is $36.18, the quartiles are $22.53 and $53.32, and the tenth and ninetieth percentiles are $15.57 and $72.30. The middle half of the sample spans $30.79, which is nearly as wide as the median itself — a reminder that an average rate applied to this catalogue is applied to parcels that differ substantially.
Band counts make the same point in a form that suits a fee discussion. Slightly more than half of the sample sits below $35, and a thin tail sits at $80 and above. A change to a flat per-parcel charge therefore affects the cheap half relatively hard, while a change to a percentage rate affects the expensive tail hard. Those are different populations, and a single headline number cannot describe a change that lands on both.
| Band | Rows | Share of 218 |
|---|---|---|
| Under $20 | 44 | 20.2% |
| $20–$34.99 | 56 | 25.7% |
| $35–$49.99 | 49 | 22.5% |
| $50–$79.99 | 54 | 24.8% |
| $80 and above | 15 | 6.9% |
What a one-point move does to a small invoice
The cheapest way to understand fee drift is to move one point and watch where it lands. The editable template this site ships uses 6% for commission and 3% for payment processing, both applied to merchandise value only, and both marked not verified. Applied to the median listing of $36.18, the two lines read $2.17 and $1.09, so $3.26 of a $36.18 order is percentage fees on the goods.
Move each rate by one point and the same lines become $2.53 and $1.45, a combined $3.98. The drift on a single median line is $0.72, which reads as noise; on a merchandise subtotal of $100 the same two points are worth $2.00, and on an order that reaches the $190.88 top of the sample they are worth roughly $3.82 by the same arithmetic. Nothing about the rate changed between those cases; only the base did.
That asymmetry is the reason a fee change should be tested against a real order total rather than against a headline. The template values used here are illustrative defaults, not any service’s published rates, and the arithmetic is shown in full below so it can be repeated with the numbers an account actually charges.
| Line | Base | Template rate | At $36.18 | At $100 | One point more |
|---|---|---|---|---|---|
| Commission | Merchandise | 6% | $2.17 | $6.00 | 7% → $2.53 / $7.00 |
| Payment processing | Merchandise | 3% | $1.09 | $3.00 | 4% → $1.45 / $4.00 |
| Combined | Merchandise | 9% | $3.26 | $9.00 | 11% → $3.98 / $11.00 |
Structural change lives in the bases, not the rates
Four base questions decide more than any rate negotiation. Does the percentage apply to merchandise only, or to merchandise plus postage? Is an inspection, a merge or a reinforcement charged per item or per parcel? Does freight price the first kilogram separately from the rest? And which volumetric divisor converts a box into billable weight? Each of these can leave every published rate untouched while moving an invoice.
The divisor is the most quietly expensive of the four. A carton measuring 40 by 30 by 20 centimetres occupies 24,000 cubic centimetres, which converts to 4.0 kilograms at a divisor of 6,000 and 4.8 kilograms at 5,000. The parcel did not change; the rule did. On a template where the first kilogram is $18 and each additional kilogram is $11, moving from one divisor to the other adds weight to a band that is priced, not a rounding difference.
The first-kilogram structure has a similar effect on small parcels. A two-kilogram parcel under that template prices at $29, of which $18 is the first kilogram, so the fixed component is more than half the line. Replacing a first-kilogram structure with a flat per-parcel charge is a structural change that can lower a heavy invoice and raise a light one in the same afternoon. Drift of that kind is invisible in a rate table and obvious in an order total.
What to watch in the next period
Four indicators are worth recording with dates, because each would show a structural change rather than a cosmetic one. First, whether the percentage base still reads merchandise only: the wording of that sentence moves more money than a point of commission. Second, whether per-item charges have been replaced by per-parcel charges, which changes the arithmetic of splitting a shipment. Third, whether the first-kilogram and continuation structure survives, or is flattened into one figure.
The fourth indicator is documentary rather than numerical: whether a dated change log appears at all. A service that publishes what changed and when can be compared against its own past; one that publishes only a current rate card cannot, and every earlier figure stays unverified by construction. That is the same standard this site applies to itself on the ledger page.
Between snapshots there is one more measurement worth keeping: the sample’s median and quartiles. They move slowly, but they move for reasons that have nothing to do with fees, so they are a check on the merchandise rather than on the seller. If the median listing drifts upward while the fee rates are unchanged, the invoice grows anyway — and that is fee drift in its least discussed form.