Method

How we reach a verdict

VerifyJersey is a small operation, and being straight about its size is part of the point. Here is exactly what we hold, what happens to your photographs, and what we will not claim.

The reference library

22 shirts, verified in person

Every reference in our library is a shirt that has been physically handled and photographed batch by batch — front, back, crest, manufacturer logo, sponsor, neck label, care label, hangtag, fabric, collar. Nothing is scraped from retail listings and nothing is added from customer submissions.

Twenty-two is a small number and we would rather state it than imply a database we don’t have. It grows every time a shirt passes through our hands. What it means in practice: when your shirt matches something we hold, the comparison is genuinely strong, and when it doesn’t, the report tells you that instead of pretending otherwise.

The check

What happens to your five photographs

In this order, every time. Several of these steps exist because an earlier version of the system got a real shirt wrong, and we would rather show the repair than hide it.

Reading a report

Why there are two numbers, not one

A single percentage hides the thing you most need to know: how much the percentage is worth. We keep them apart deliberately.

Authenticity estimate

1–99

How genuine the shirt appears on the evidence available. Never 0 and never 100, because photographs cannot deliver certainty in either direction and a number that implies they can is dishonest.

Evidence confidence

Very low → High

How much the first number is worth. A high estimate with very low confidence means the shirt looked fine and we could barely see anything — that is a different situation from a high estimate backed by sharp photos and a matching reference, and you deserve to be able to tell them apart.

Reference match

None → Exact

Whether we actually had a verified shirt to compare yours against, and how close it was. If this says None, the report is a reasoned opinion about the shirt in isolation — worth having, but not the same as a comparison.

Limits

Five things we will not do

We never invent a comparison

If no verified reference was supplied to the analysis, the report cannot claim one. This is enforced by our own code after the fact, not left to good intentions.

We never issue a certificate of authenticity

We are not in a position to guarantee origin from photographs, so we don't sell a document that implies we can.

A customer submission never becomes a reference

Shirts only enter the library after being handled and photographed in person. Nothing you send us is silently promoted into the evidence used on somebody else's shirt.

Seller feedback carries no weight

Marketplace reputation describes a seller's history across many transactions. It says nothing verifiable about the individual garment, so it is not treated as evidence in any finding.

'Replica' is never used to mean counterfeit

Replica is the licensed retail tier that most fans own. Counterfeit means unlicensed. Conflating them frightens people out of perfectly genuine shirts.

The largest limit is the one no method can remove: photographs can find contradictions, but they cannot establish where a garment was made. A shirt that survives every check might be genuine, or might be a very good counterfeit. When a case reaches that point we record it as unresolved — here is one we could not settle, published in full.

See the method on a real shirt

Two shirts examined in public, photograph by photograph — or send us yours.