The reading protocol: four steps, and what each warning banner checks

Step 1 — Place the order on the timeline

Read the stage before you read anything else, because the stage decides which actions still exist. Five states cover almost every case, and the state you are in is visible from your own screens.

1. A link, nothing paid. Nothing has been bought for you yet, so changing your mind is a withdrawal rather than a dispute, and it costs nothing. Read Stop 2 — handing the link over and the cancel window, question by question.
2. The item is paid for and travelling to the warehouse. The item money is committed. What stays open is correcting the address, the size or the variant, plus the arrival record itself. See Stop 3 — what closes when you pay.
3. Booked in and sitting in storage. Two clocks run at once: the free storage days and the photo window. See Stop 4 — the warehouse clock and Stop 5 — reading the photos in time.
4. Packed, freight not yet paid. Merging is still possible and the chargeable weight has not been fixed. See Stop 6 — merge, repack, or ship as is and the merge planner.
5. Shipped. The levers left are the customs line, the address the carrier holds, and the claim window that opens on delivery. See Stop 7 and Stop 8.

Step 2 — Name the action you want to take

Two people at the same stage often want opposite things, and the page that helps depends on the verb, not on the parcel. Match your verb to one of the five lines below.

1. Cancel or withdraw. Time-critical, and only available in the earliest stage. The cancel window and Stop 3.
2. Change an address, a size or a variant. Fields lock in a fixed order, and the first one to lock is rarely the one people expect. Address and declaration.
3. Ask for another photo or a different check. A photo request is a specific list, not a polite sentence. The coverage scorecard and Stop 5.
4. Merge, repack or ship. The decision with the largest arithmetic effect on the bill. Merge planner and three merge timings, three totals.
5. Claim for damage or a wrong item. The claim lives on evidence you kept, not on how annoyed you are. Stop 8 and return versus refund.

Every branch above ends inside this site, because the method should not depend on anyone’s goodwill. The rows those checks act on sit on the destination channel, so run the same check there before you commit: a check you can repeat is worth more than a conclusion you have to trust.

Run the same check on Kakox

Outbound links to Kakox are tracked so the destination can attribute the visit. They do not change what this page says, and nothing here is sold by us.

Step 3 — Weigh how much is already committed

The amount at stake decides how much effort the next check deserves. Four bands cover the cases readers write to us about, and each band has one tool that does the arithmetic in full view.

1. One item, a small amount. A single waterfall pass is enough: eight lines, and the point is to see which two dominate the total. Cost waterfall.
2. Several items paid, nothing merged. Merging changes weight and volume at the same time, so decide before freight is priced. Merge planner.
3. A parcel that will be billed on size. Enter the three sides and the real weight; the larger of the two decides. Chargeable weight.
4. Resale, where the total has to leave room. Landed cost first, break-even sale price second. Margin tool.

Four banners, and what each one is actually checking

A banner is not a fault report, and it is not a wall. It is a check that stopped, and each kind of check names something you can look at. We do not hold the destination’s internal rule list, so what follows is what each banner correlates with on the public pages, plus the handling that does not make it worse. The wording on your own screen is the authority; if it names a field, that field is what is being compared.

Step 4 — Add the destination

The last step changes which lines exist, which fields are required, and how much of the bill is tax rather than carriage. Four destination groups cover the readers this runbook is written for.

1. United Kingdom. Declaration values and the handling line matter more than the carrier name. Reading the customs line.
2. United States. Bands, thresholds and the paperwork that travels with the parcel. The same customs reading and the waterfall.
3. Canada and Australia. Long-haul lines and seasonal cut-offs decide more than the headline rate. Cut-off dates that move.
4. Elsewhere. Start from the charging basis rather than the quote. Chargeable weight and ranking lines by what they charge on.

Six known limits of this protocol

Limits are published because a method without them reads like a guarantee. Each line below is something this protocol cannot do, stated before you rely on it.

1. It cannot see your account. Every step works from public pages and from what you can read on your own screens. Where a status lives only inside an account, we describe the shape of the state and not its current value.
2. It does not hold the destination’s rule list. Banner families are described by what they correlate with. If the destination changes what a banner means, this page is wrong until we re-check it.
3. Windows are described behaviourally. We name the action that stops being available, not a published deadline, because the deadlines we could find were not dated on a public page.
4. One channel, eight pages. The sample behind the site is 192 rows from eight pages of one channel. The snapshot shows roughly 8,424 rows per channel, so the sample cannot stand in for the whole list.
5. Labels are labels. Category and brand come from reading titles. On the snapshot, 26 rows stay unclassified and 56 read as brand not readable; both are reported as they are rather than guessed.
6. It returns a page, not a verdict. No step outputs a score, a percentage or an “is this safe” answer. The output of a tree is the next page to read.

Four things this site refuses to do

1. No link survival-rate promises. Publishing a percentage would require re-crawling every listed row on a fixed schedule. We do not do that, so any figure we printed would be invented rather than measured.
2. No link parameter decoder. We cannot verify how a third party parses its own addresses, and a decoder we cannot test is a guess wearing a user interface. The five-step recovery for a dead link does the job with checks you can see.
3. No bypass steps for banners. A banner is a check, not an error state, and instructions to slip past a check would be instructions to break a rule we cannot read.
4. No prices, stock states or channel rankings. Prices and availability move faster than a weekly review can honestly follow, and a ranking would need a measurement we refuse to fake.

Where each step sends you

The four steps above point at twenty pages. If you want the shortest path through them, use the mapping below, then check any count you meet against the page that shows its working.

• The eight order stops — the timeline the first step reads from.
• The reading protocol tool — the same four steps applied to one row, with a three-state result.
• Warning banners — the long version of the four families above.
• Sampling receipts — where the counts in Step 3 and the limits above come from.
• Twenty-eight questions — for readers who want the answer before the method.
• The masthead — who signs this, and how to dispute a line in it.