Line-by-line: matching a parcel to a carrier instead of a habit

Any ranking of services is meaningless until the parcel and the priorities are fixed, because the answer changes with both. We therefore score against a reference parcel: about six kilograms chargeable, outer dimensions in the region of 45 by 35 by 25 centimetres, mixed apparel with one pair of sneakers, no battery and no liquid, destination United Kingdom, declared value in the low hundreds. Change any of those fields and you are running a different comparison, which is the point rather than a caveat.

Eight dimensions carry the score, and we publish the weights so they can be argued with. Cost band on the per-kilogram line takes 25. The transit range, taken from community-reported windows rather than from a service promise, takes 15. The volumetric divisor together with the rounding step takes 15. Acceptance rules for restricted goods take 10. Weight and size ceilings take 10. Peak behaviour, meaning cut-off frequency under load, takes 10. Tracking granularity and handover visibility take 8. Declared-value ceiling and claim process take 7.

The weights encode one opinion: cost and measured transit matter most, because they are the two things a buyer can check afterwards. The divisor earns a weight that surprises people until they run their own box through it, at which point a lenient rounding step and a strict one can differ by a band on the same parcel. Acceptance rules get a moderate weight because when they bind they bind absolutely: a service that refuses what is in your parcel does not score badly, it scores nothing.

Using the sheet takes ten minutes. Score the services your agent actually offers on your route, change the weights to match what you care about, and re-run it. Someone with a hard delivery date should raise transit and peak behaviour and drop the cost band; someone shipping a single expensive item should raise the declared-value ceiling. Our order is not the finding here. The finding is that the order is derived from stated weights rather than from the last service that happened to work.

No figure for any service appears in this article, and that is a deliberate limit rather than an oversight. Rates are quoted at payment time from tables that move, and a number written on a page is wrong within a quarter while looking authoritative for a year. What the ranking records is shape: which kind of service leads on which dimension, and how the leader changes when the parcel changes.

Four service archetypes do most of the work. Express-integrated carriers run their own end-to-end networks and lead on tracking granularity, declared-value ceiling and claim handling. Route-specific economy lines buy space on consolidated builds for a single corridor and lead on the cost band for mid-weight parcels, with thinner tracking. Postal-style economy services lead at very low weight and trail on visibility, cut-off frequency and peak behaviour. Specialist lines accept one goods class and win outright whenever your parcel is in that class.

The mechanism behind each leader is structural rather than commercial. Cost is set by how the service buys space, so consolidated builds beat networks on mid-weight parcels where the fixed cost of a network cannot be spread. Transit is set by cut-off frequency and first-leg access. Tracking granularity is set by how many handovers the service owns and therefore how many it can report. Because those are properties of a business model rather than of a brand, the ordering stays broadly stable across destinations while flipping with parcel weight and contents.

For the reference parcel, first position goes to a route-specific economy line. It wins the cost band by a margin the other dimensions cannot close, it publishes a divisor that does not punish a box of this size, and its weak dimension, tracking granularity, costs nothing on a parcel the buyer is not watching hourly. Its exposure is peak behaviour: fewer cut-offs a week means a missed build costs days rather than hours, which is why the ranking does not survive into December unchanged.

Second position goes to an express-integrated carrier. It takes tracking granularity, declared-value ceiling and claim process outright, and it loses the cost band at six kilograms. The interesting part is how quickly it takes first place as the parcel changes: below roughly two kilograms the fixed per-parcel lines dominate and the express service becomes competitive, and above a declared-value ceiling that the economy line will not cover, the express service becomes the only practical choice regardless of cost.

Third position goes to a second route-specific line that earns its place on acceptance rather than on score. It takes the restricted-goods dimension for the corridor, so the moment the parcel contains an installed battery or a sealed liquid it moves to first. Worth noting is that this is not the same ranking with a different winner. When acceptance rules bind, the eligible set is smaller, and a ranking of four services becomes a ranking of two, which is a different comparison rather than a reshuffled one.

The shift with weight is not gradual, which is why a single ranking for all parcels misleads. Below about two kilograms the fixed per-parcel lines dominate the total and the express-integrated service becomes competitive on cost as well as on tracking. Between two and ten kilograms the per-kilogram band dominates and a route-specific line usually leads. Above ten kilograms the field compresses for a different reason: ceilings and acceptance rules remove services from the eligible set, so the comparison is between fewer options and the winner is sometimes determined by who will carry the parcel rather than by cost.

The middle of the table is where composite scores bunch. Positions four to eight in our sheet sit within a few points of each other, and inspection shows they differ from one another on a single dimension, usually tracking granularity or cut-off frequency. A composite score difference that comes from one field is a difference you can decide directly, without the score.

So at these positions we stop ranking and start eliminating. Three questions do the work: will the service accept what is in the parcel, does the parcel fit under the weight and size ceiling, and how many cut-offs a week is the service running in the month you are shipping. A service sitting four points lower on the composite can be the only one that answers all three, and at that point the ranking is decorative rather than decisive.

Being low in our table means narrow rather than poor. Several entries sit in the bottom half because they serve one destination well, or one goods class, and score middling on dimensions they do not compete on. For the parcel that matches their specialism they frequently hold first place. Our weights make them look average because the weights assume a general parcel going to a general place, which is exactly the assumption a specialist line is designed to break. Telling whether you are that case takes two checks and no scoring: whether the goods class you are shipping appears on the service restriction list as something it handles rather than something it excludes, and whether your destination is one the service names. When both answers are yes, the composite score is irrelevant, and the only remaining question is the cut-off frequency in the month you are shipping.

It cannot tell you what your parcel will cost. The rate is quoted at payment time from a table we do not publish, and the list of services your agent offers on your route is account-specific and changes with seasonal volume and capacity that we have no visibility into. Anyone quoting you a per-kilogram figure for a named service on a general page is either republishing an old table or guessing.

It cannot tell you what will happen to your parcel. Last-mile partners vary by region within a single destination country, remote-area surcharges appear after the quote rather than in it, and customs behaviour is a property of the destination rather than of the line. A transit range describes a distribution of many parcels, and any single parcel samples that distribution once. Reading the upper bound as your likely outcome is pessimism; reading the lower bound as it is a mistake.

It cannot tell you what December looks like. The ordering that matters in a peak month is set by cut-off frequency and space allocation, both of which move, and both of which are rarely published. That is why the sheet ends with two live checks rather than a conclusion: ask how many cut-offs the service is running this month, and confirm the acceptance list for your goods on the day you pay. The habit worth dropping is reusing the same line because it worked once. The habit worth keeping is re-running eight dimensions against the parcel in front of you.

One maintenance note belongs here, because a ranking decays in a specific way. It starts being treated as a property of the brand rather than a property of the parcel, and from that point it stops being re-derived. When a service changes its divisor, narrows its acceptance list or moves to a different first leg, its row changes even though nothing about the brand did, and a reader working from memory will not notice. Our sheet carries a date for that reason and gets re-run each quarter. Re-score when an input moves, not when the calendar reminds you to.