One tag in a batch reads much worse than the others
Either it was damaged after leaving the factory, or it is not the same product as its neighbours. Reading it and comparing the reported chip and tag type against a few others separates those two in seconds, and they lead to opposite conclusions.
Two possibilities
The tag was damaged: creased in transit, bent during handling, or manufactured with a weak bond between chip and antenna.
Or the tag is a different product: a mixed delivery where part of the batch is another reference, another antenna size, or another tag type.
Read it and read five neighbours. If the reported chip, tag type and free space all match, it is damage. If they differ, it is a mixed delivery.
This distinction matters because one is a single tag to discard and the other is a supplier conversation about the whole order.
Damage after manufacture
A partial break or a fatigued bond between chip and antenna reduces what the tag can collect without stopping it entirely. The result is a tag that reads only at contact while its neighbours read at a centimetre.
Nothing is visible from outside. Holding a thin label up to a light occasionally shows a crease across the foil, and usually shows nothing.
A tag in this state does not recover and tends to get worse. Discard it rather than deploying it somewhere it will fail later.
One weak tag in a box is ordinary. Several is a transport or handling problem worth tracing.
A mixed delivery
Batches get combined at the supplier, and the packaging rarely changes when that happens.
A different antenna size explains a different read distance completely, with no fault anywhere. A different tag type is more serious, since part of your deployment may not read on some phones at all.
Check enough tags to know the proportion, taking them from different parts of the delivery rather than the top of one stack.
Then decide whether the substitute is acceptable, and if it is, record which tags are which. An undocumented mixed deployment is difficult to maintain.
What to do about it
Never install a tag that reads worse than its neighbours. It is the tag that will fail first and it will fail where reaching it costs the most.
Keep a known good tag from each batch, unwritten and labelled, as your comparison reference.
Verify every tag before installation, not a sample, wherever reaching the tag again is expensive. Reading each one on a bench costs seconds and a return visit does not.
And record the incoming failure rate per batch. Two numbers over two deliveries tell you whether one bad tag was ordinary or the start of something.
Frequently asked questions
Is one bad tag in a hundred normal?
For a physical product, a small number of failures is ordinary. What matters is whether the rate is stable across deliveries.
Could the weak tag just be a different chip?
Yes, and that is why you read it rather than testing it. A different reference with a different antenna explains a different distance without any fault.
Can a weak tag be improved?
No. Antenna damage and a fatigued chip bond are both permanent, and a tag costs cents against the labour of installing it twice.
Should I return one weak tag?
Not usually worth the effort. A pattern across a delivery is worth raising, which is why counting matters more than reacting to one.