How a claim gets onto this site
The test a claim has to pass
Five questions, asked in order. A claim that fails any of them is rewritten or dropped rather than softened with a hedge.
Can a reader check it without asking us?
If the only source is this site, the claim does not belong on it. Everything published should be verifiable somewhere a reader can reach independently — a seller listing, an agent's own page, a dated reading.
Does it carry a date?
Prices, fees, counts and policies move. A figure without a reading date cannot be checked later, only believed now, which is the opposite of what a reference is for.
Is it a fact or a judgement?
Facts are stated flatly. Judgements are labelled as judgements and given the reason. Mixing the two is how a site ends up sounding confident about things it cannot know.
Would it still be true if the reader were sceptical?
Claims that only work on a trusting reader are marketing. The test is whether the sentence survives being read by someone actively looking for the trick.
Does it require an experience we have not had?
See the review scope table.
What counts as evidence here
Five kinds of source, ordered by how much weight they carry, with the reason for the ordering.
- A dated reading of a public API
- Reproducible, machine-readable and timestamped. It cannot be argued with, only superseded.Where to look: Every count and price on this site.
- A page the operator publishes about itself
- Authoritative about its own rules, and silent about outcomes. Useful for mechanics, useless for quality.Where to look: Fee structures, route tables, QC process descriptions.
- A listing that can be opened and checked
- Verifiable at the source, and perishable — which is why the reading date travels with it.Where to look: Every recorded product row.
- A pattern across many listings
- Stronger than one example and weaker than a rule, because a catalogue can be wrong in the same way many times.Where to look: Brand-field coverage, variant counts, category spread.
- An anecdote, however detailed
- Not evidence. It is a hypothesis about what to go and check, and it is labelled as such when it appears at all.Where to look: Almost nowhere on this site, deliberately.
Read top to bottom, the list is also a hierarchy of perishability: the API reading is the most durable, the anecdote the least, and the ordering is why the site leans on the first three.
Claims that failed the test
Five things this site could have said and does not. Each is a common claim in this market, and each fails a different question above.
| Claim that could have been made | Why it was rejected |
|---|---|
| "Oopbuy is the cheapest agent" | No stable answer exists across weights, routes and destinations, and any single ranking would be wrong for most readers |
| "Points convert at approximately X per unit" | The conversion is the operator’s to change, so any figure is stale on the day it is published |
| "QC photos catch problems" | A photo documents; it does not judge. The stronger and true version is that photos are visual only |
| "Our readers report fast support" | Depends on anecdotes this site has not collected reproducibly, and would imply a survey that was never run |
| "This spreadsheet has 10,000 verified finds" | A verification claim with no verification behind it — the exact pattern this site exists to document |
Verification questions
Four answers about how claims are handled.