Weighting Five Criteria When You Choose an Agent in 2026

The five criteria, and what each one measures

Billed-weight basis is the first criterion because it sets the unit that everything else is priced in. It covers three parameters: the divisor used to convert volume into weight, the per-parcel minimum below which weight stops mattering, and the rounding rule applied at the end. A divisor difference between 5000 and 6000 g/L is worth about 20% on a bulky box — 8.00 kg against 6.67 kg on 40,000 cm³ — which is larger than most fee differences between services.

Fee transparency measures whether the stack can be reconstructed before payment. It covers the base a percentage is applied to, per-order and per-item minimums, payment surcharges, and whether value-added operations such as extra photographs or reinforcement are itemised. This site’s 218-listing snapshot taken on 2026-09-29 puts the median list price at $36.18, with 44 rows below $20.00 and 56 more in the $20–$34.99 band; on goods in those bands a minimum charge can outweigh any headline percentage.

Warehouse clock measures the free storage window granted from intake, what happens when it closes, and whether a merge resets it. Intake evidence measures the photograph set a buyer receives, the angles covered, and whether a reshoot can be requested before dispatch. Recoverability measures the least glamorous thing on the list: whether a problem produces a written answer, and whether a return is possible at all once a parcel has been consolidated.

Weights: an editorial choice that decides the order

No weighting is neutral. The three profiles below are stated so that a reader can see where an order comes from rather than inferring it, and so that a reader whose situation differs can substitute their own column. The balanced profile suits a first order of ordinary clothing; the bulky-goods profile suits jackets, accessories and anything that ships in a large box; the high-value profile suits carts where the goods are worth several times the postage.

The weighted board, and how one candidate is scored

Scoring runs on a four-point scale per criterion, multiplied by the weight and divided by four, so a candidate that scores full marks everywhere reaches 100. Two rules keep the board from manufacturing precision. A criterion is scored only from a parameter the service publishes, and an unpublished parameter is left unscored rather than filled with a middle value; and the effect of an unscored criterion is stated rather than hidden, since a missing criterion at weight 30 caps the verifiable total at 70.

The worked board below uses two illustrative candidates rather than two named services, because the parameters that would be needed to score a real service — its divisor, its minimum, its storage window — were not verified for this cycle. The arithmetic is the part worth reproducing.

• Under the balanced weights, candidate A leads by 5.00 points, entirely on fee transparency.
• Under the bulky-goods weights the same scores produce a tie at 82.50, because billed weight and intake evidence both gain weight.
• Under the high-value weights the gap narrows to 3.75 points, as evidence and recoverability take 55 points between them.

Entry by entry: the billed-weight basis, ranked first

This criterion ranks first in two of the three profiles for a simple reason: it multiplies every kilogram that follows. A buyer who knows the divisor and the minimum can compute the billed weight of a parcel before it exists, and a buyer who does not is choosing between services on the basis of a rate that may never be applied to their goods.

The questions that score it are narrow enough to be answered in a message. Which divisor is used? Is there a per-parcel minimum, and does it apply after a merge? How is a part-kilogram rounded? A service that answers all three scores full marks on the parameter; a service that answers none leaves the criterion unscored and caps its verifiable total regardless of how attractive its other numbers look.

The consequence of getting the divisor wrong is concrete rather than theoretical. On the 40,000 cm³ box used in the worked table, the difference between the two common divisors is 1.33 kg of billed weight, which is the same order of magnitude as the difference between a cheap line and an expensive one on a small parcel.

Entry by entry: fee transparency and the warehouse clock

Fee transparency ranks second under the balanced weights and drops under both specialised profiles, because it is a criterion about predictability more than about price. Its highest-value question is not the percentage but the base: a fee applied to the item price behaves differently from one applied to the item price plus domestic freight, and a per-item minimum changes the answer for every cheap row in a cart.

The warehouse clock ranks consistently in the middle, and it punishes two different buyer behaviours in opposite ways. A buyer who orders in batches uses the free window well and rarely pays storage; a buyer who waits for a slow merge partner converts a saving on postage into a storage charge, which is the trade described in the note on consolidation and billed weight.

Both criteria share a property that makes them hard to score from memory: their parameters change. Terms stated in one season are not guaranteed in the next, and a score recorded without a date is a score that may already be wrong. Every cell in a board like this should carry the date it was checked.

Entry by entry: intake evidence and recoverability

Intake evidence carries the largest weight under the high-value profile, where a defect found after dispatch costs more than the entire fee stack on a cheap cart. The parameters are countable: how many photographs arrive by default, whether specific angles can be requested, whether a measurement can be asked for, and whether a second set is available before the parcel is merged.

Recoverability sits last under two profiles and second under the third, which reflects how rarely it is used and how expensive it is when it is. The evidence for scoring it is partly negative: in this site’s catalogue snapshot, 52 of 218 rows carried no short note at all, and a service that leaves questions unanswered in public tends to leave them unanswered in private as well. A written reply that names an outcome is the parameter that matters most, because it is the only one that survives into a later dispute.

Taken together, the last two criteria are the ones that separate candidates whose headline numbers look identical. Neither is visible in a rate table, and both are cheap to check before a first order is placed.

Objections and disputes

The strongest objection to a weighted board is that the weights are a choice, and the objection is correct. Reordering the columns above by a few points is enough to change a five-point lead into a tie, which is why the weights are published rather than implied; a board that hides its weights is presenting a preference as a measurement.

A second objection concerns disclosure bias. Services that publish more detail accumulate verifiable scores while services that publish less accumulate blanks, and blanks are not evidence of poor practice. The rule that an unscored criterion caps the total is a mitigation, not a solution: it penalises silence without pretending that silence is misconduct.

A third objection is compression. A single number cannot hold a conditional clause — for example, a minimum that applies per parcel before a merge but not after it — and collapsing such clauses into a score loses exactly the information a careful buyer wanted. The board is a way to narrow a candidate list, not a way to finish the decision.

The final dispute is about time. Terms change within a single year, and a table assembled in 2026 will describe 2027 badly unless its dates travel with it. Scores that carry no check date should be treated as unverified, including any score reprinted from a page that does not show one. A buyer who keeps a dated board of their own ends up with something no published comparison can supply: a record of which parameters moved, and when.