Reading market ratings and review scores without being fooled

Who actually leaves feedback

Feedback is not a random sample. It comes disproportionately from people whose order went unusually well or unusually badly, while the ordinary majority say nothing. That gives most rating distributions a shape with a heavy top, a small tail and almost nothing in between, and it means an average score is describing the loud edges rather than the typical experience.

Directory rankings and placement

A separate distortion applies to lists ranking whole markets. Placement on many directories is purchasable, and where it is not purchasable it is often driven by affiliate arrangements. Neither is disclosed as a matter of course. A market appearing first on several lists at once may be the best on some measure, or may simply have the largest promotion budget, and from the outside the two are indistinguishable.

What to read instead of the number

• The recency of positive feedback rather than the count, since an account can coast on old reputation
• How the seller answered negative feedback, which shows more than a wall of high scores
• Whether ratings survive address rotation, since a system that resets teaches nothing
• The dispute record where visible, because that is where behaviour under stress appears

On this library own position

The same scepticism applies here. This site publishes verified Nexus addresses, which is a relationship worth knowing about when reading anything it says about Nexus. The defence against that is method rather than assurance: criteria are stated in advance, gaps are marked as gaps, and a reader can apply the same criteria to reach a different conclusion.