Rule one: a claim is read from records, not from the conversation
The first rule is not printed on any agent page because it does not need to be. A claim is decided from the records attached to the order, and a chat message is not a record. The order line, the intake record, the photo set, the weight and dimensions captured at the warehouse, and the tracking history are records. What you wrote in a message is context, and context explains a claim without ever deciding one.
In practice this means the claim is assembled before you write anything. If the photo set does not show the defect, no amount of later wording substitutes for it. If the weight recorded at intake disagrees with the weight on the shipping label, that single pair of numbers settles a missing-part question faster than a paragraph describing how the parcel felt when it arrived.
The common misreading is that a helpful reply is a decision. An agent writing that they will look into it has said nothing about the outcome and has not changed the file. The window keeps running while the file stays still, and the reply that felt like progress is the reason nobody noticed the deadline.
Ask which documents are attached to your order before you argue about them. A record that exists can be pointed at; a record that does not exist cannot be created afterwards, whatever the explanation was.
Records also have authors, and that matters when two of them disagree. The intake record is written by the warehouse, the tracking history is written by carriers, and the order line is written by you at the moment of purchase. A contradiction between two of those three is a much stronger position than a contradiction between any of them and a message, because the reader can check the documents without having to trust anybody. The order line is the only one of the three you author, which is why it is worth writing precisely at the moment of purchase instead of reconstructing it from memory later.
Rule two: the listing text is part of the standard you agreed to
Whatever the seller wrote in the listing forms part of what you agreed to buy, and that cuts in both directions. A flaw the listing described in words is not a mismatch, because the described product is the one you chose. A specification the listing stated clearly is a standard against which the delivered item can be measured without anybody’s opinion being required.
The operational consequence is to keep the listing text as it stood on the day you ordered. Listings get edited, variants get replaced, and the page you can still open next month may not be the page you agreed to. If the difference between those two versions matters to the claim, only one of them is worth quoting.
The misreading here is treating the gallery as the specification. On most listings the photographs are illustrative and the variant you selected plus the written description are the parts that can be quoted back. A claim built on a gallery image invites a reply about photography, and that reply is not evasive even when it feels that way.
Read the description once with a pen in hand and mark the sentences that state a fact rather than a mood. A sentence naming a material, a measurement or a colour is quotable, while a sentence about quality or value is not, and a claim should be built only from the first kind. Most listings contain two or three such sentences, and those are the whole of the standard you can hold anybody to. Everything else in a listing is atmosphere, and atmosphere is not quotable.
Rule three: a merged parcel is judged through its packing record
Once items are consolidated, the parcel becomes the unit. Individual items inside it are no longer separately measurable, and a question about one item among five has to be answered from the intake record plus the packing record rather than from a fresh look at the contents.
That makes the intake record the durable document. It is written while the item sits alone on a bench, and it is the only moment at which the item had its own photographs, its own weight and its own dimensions. If you expect to raise anything about a single item inside a merged parcel, the record has to exist before the [merge and pack](/steps/merge-parcel/) step is submitted, not after.
Flow dependency: evidence is captured per stage. The intake record exists between arrival and merge, and after the merge the same item has no separate measurement, so a later claim rests on the earlier document.
There is a second consequence that catches people out. A consolidated parcel has one weight and one set of dimensions, so a claim about a single item inside it cannot be argued from the parcel number, and asking for the parcel to be re-weighed proves nothing about the item. What proves something is the intake record for that item, written days earlier, when it was the only thing on the bench. Timing is therefore part of the evidence rather than a separate administrative detail.
Rule four: each instrument runs its own clock
There is no single deadline on an order. Each instrument carries its own, they start at different moments, and they end at different moments. Treating the order as one countdown is how a buyer arrives at the right complaint with the wrong instrument already expired.
| Instrument | Clock starts | Clock governs | Date lives in |
|---|---|---|---|
| Return to seller | Item intaken | How long a return is worthwhile | Storage terms, not here |
| Refund on item | Photo set produced | Raising a mismatch before sealing | Your own order state |
| Transit claim | Delivery or a bad scan | Reporting damage or loss | The carrier form |
| Customs query | A request from clearance | Supplying a document | The notice itself |
Two of those dates are in documents you were handed, one is visible in your own account, and one belongs to terms this page does not restate. Where a figure would normally sit, the honest entry is that no published number applies, and the question goes to the agent with the reply kept.
The practical habit is to write the four dates into the order notes the day the order is placed, leaving the two unknown fields marked as unknown. A field that says unknown can be filled in later when an answer arrives, while a field left blank is indistinguishable from a field nobody remembered to check, and three months later you will not be able to tell which of the two you are looking at. Two of the four dates are usually known on the day of purchase, the other two become knowable at intake, and intake is the moment to write them down.
Misreadings that sink claims which were otherwise sound
Most claims that fail were reasonable complaints. They failed on handling rather than on merit, and the same handful of errors accounts for nearly all of them.
- Raising the defect after the merge, when the evidence that would have settled it was captured before.
- Describing the wrong item instead of quoting the code on the label, which turns a fact into an opinion.
- Writing the whole history into the first message, so the one sentence that matters is read last.
- Treating a reply as a decision, and letting the window close while waiting for one that never came.
- Sending the item back before a return has been agreed, which produces a parcel nobody is expecting.
Each of the five has the same shape: an action taken at the wrong stage, or a document that was available and was not used. None of them requires a better argument, and all of them are avoidable by reading the record before writing the message.
There is a sixth error that outranks the rest, and it is waiting. Every instrument in this process is governed by a date that was fixed before you noticed the problem, and none of those dates moves because a conversation is still open. A claim written on the last day with one clear sentence does better than a claim written earlier without the frame reference that would have settled it. Speed is not the virtue here; sequence is.
Open the order state, the intake record and the photo set in that order before drafting anything. The [returns topic](/table/returns/) and the [arrival stage](/steps/warehouse-in/) both describe that sequence from different ends of the same process.
Flow dependency: per-item evidence exists only between intake and merge. While an item is stored alone it has its own photographs, weight and dimensions; once it joins a consolidated parcel that record is the only measurement that will ever exist for it, so a later claim rests on a document written before the claim was imagined. (Flow dependency)
This note explains the mechanics; the list is where the items are.