When is cryptographic authentication worth its cost?
When a convincing counterfeit would cost more than the scheme does, counting all of it. The chips are the visible cost and the smallest one. The verification service, the key management, and keeping both running for the life of the product are where the money actually goes.
The cost lines
| Line | Usually costed? |
|---|---|
| The chips | Always |
| Programming keys during production | Sometimes |
| The verification service | Rarely |
| Key management and its recovery | Rarely |
| Running both for the product's life | Almost never |
| Deciding what a failure means | Almost never |
The first line is the one quoted per thousand and the one every proposal leads with. The last four decide whether the programme is affordable.
When it is worth it
Where a counterfeit causes harm rather than lost margin: a part in a machine, something consumed, something whose failure lands on you.
Where the item is valuable enough that a fake is worth manufacturing, and where the counterfeiters are competent enough to reproduce everything else about it.
Where the tag grants something. A ticket, a membership or an entitlement is defeated entirely by duplication, and nothing cheaper prevents it.
And where you will still be running the verifier in ten years, because otherwise the scheme expires before the goods do.
When it is not
Where the tag opens a public page. A copy gains a page anybody could already reach.
Where the real problem is a tag being rewritten rather than duplicated, which locking addresses at no cost.
Where nobody will check. Verification that nobody performs protects nothing, however sound the chip.
And where the answer to a failed check has not been decided, since an alarm with no response is a cost with no benefit.
Deciding without guessing
Write down what a convincing fake would cost you in a year. One figure, defensible, including the harm rather than only the lost sale.
Price the six lines above over the same period, asking the supplier specifically about the service and the key management rather than the chips.
Compare. If the figures are close, the answer is no, because programmes overrun and counterfeits are patient.
And consider the alternative first: a system arranged so that a duplicate is detectable by its behaviour costs far less and catches a surprising amount.
Frequently asked questions
What does it really cost?
Far more than the chips. The verification service, the key management, and running both for the life of the product are the larger lines.
When is it clearly justified?
When a counterfeit causes harm rather than lost margin, or when the tag grants an entitlement that duplication defeats.
When is it clearly not?
When the tag opens a public page, when the real risk is rewriting, or when nobody will actually check.
Is there a cheaper alternative?
Often. A system that detects duplicates by their behaviour, such as the same tag appearing in two places, catches a great deal for very little.