How the ranking is made
A ranked table implies precision it does not have. This page is the correction, and it is deliberately the first thing in the reference section.
The rule behind the order
Expected cost: roughly how often something happens, multiplied by roughly how much it takes when it does. Nothing more sophisticated, and nothing numeric, because a number would imply a measurement that does not exist.
| Band | What it means here |
|---|---|
| Very high | Common and takes a large share of what you hold |
| High | Either common with a medium bite, or uncommon and total |
| Medium | Common but bounded, usually to one order |
| Total | Rare, and takes everything in scope when it lands |
| Unbounded | No money moves at all, and it cannot be reversed |
Why there is no dataset
Nobody publishes reliable figures for this. Platforms do not, for obvious reasons. Surveys reach the people willing to answer surveys, which is not the same population. Forum reports are heavily skewed toward the dramatic, because a routine loss is embarrassing and an exotic one is a story.
So anybody quoting precise percentages here has invented them, and the honest alternative is to rank rather than to measure, and to show the reasoning on every page.
Where the order is least certain
- Ranks 1 to 3 are firm. All three are common, cheap for the other side, and expensive for you. The order between them is arguable, their being at the top is not.
- Ranks 7 and 9 are the least certain. Both are rare and total, and rarity is exactly the quantity nobody can pin down. A reader with a compromised machine would reasonably put rank 9 far higher.
- Rank 12 is deliberately uncomfortable. It costs no money, so by the stated rule it belongs last, and it is the only entry that cannot be undone. The rule was not bent to make it look more important.
What would change the table
| Change | Effect |
|---|---|
| A mechanism on the platform changes | The affected entry is rewritten |
| An approach becomes more common | Its rank moves and the reasoning is restated, not quietly edited |
| Somebody points at a cost not accounted for | The entry changes |
| Somebody disagrees with a band | Only moves if they can name the step they disagree with |
| Somebody offers to pay for a placement | No, and the offer is declined |
What is deliberately absent
No availability figure, because this site measures nothing. No ratings or reviews, because there are no customers to quote. No advertising or affiliate arrangements, and no money from any platform. Nothing about products, sourcing or shipping. The modification date in the markup follows real edits rather than the current day, because a date that advances daily while nothing changes is a false signal.
Why ranking beats scoring
A score implies a scale, and a scale implies units. There are none here. A rank implies only that one thing is placed above another, which is a claim that can be defended or attacked on its own terms.
It also fails more honestly. If the ordering is wrong, somebody can say which pair is swapped and why. If a score of 7.4 is wrong, there is nothing to argue with, which is precisely why scores are popular with sites that would rather not be argued with.
What was considered and left out
| Excluded | Reason |
|---|---|
| Seizure and prosecution | Nothing a reader changes by behaving differently, and this list sorts actions |
| A market ending | Same reason, and it is folded into the balance entry where a reader does have a choice |
| Network level surveillance | Real, and not something twelve rows of ranked advice improve |
| Product quality and harm | Outside the boundary this site keeps |
How to argue with the table
Name the entry, name the step in its reasoning you reject, and say what the consequence for its position is. That is a change that can be made. Saying the order feels wrong is not actionable, and neither is asserting a frequency without a basis, since the whole page above explains why nobody has one.