Abstract: The need of confidence that a product or service actually possesses declared behavioural / structural characteristics is the main reason for certification. It seems that this very concept and its implications have not been discussed deeply enough in literature: mostly it was insisted in describing tests and measures [Voa00] [Tri02] [MTT09], and little attention was paid to the 'confidence' aspect itself. This is probably one reason why certification was often mistaken by users as a guarantee, and by producers as particularly severe verifications and validations. If we consider the software, then the confidence is generally weakened, and it can be observed that responsibility disclaimers associated to products are much more frequent than certificates. Even the term 'certification' was somewhat banished in the US for years, for the unspoken threat that customers unions could claim refunds for unsatisfying 'certified' services.