Quantum testing, validation and certification: what is the difference?

Testing records performance under stated conditions. Validation asks whether evidence supports a specified intended use. Certification adds a conformity statement under an identifiable scheme and authority. For quantum procurement, the value of any document depends on its scope, method, independence and acceptance by the relevant buyer; the labels alone do not establish readiness.

Testing a defined device and configuration

A vendor can truthfully say that a device was tested while leaving the buyer’s central question unanswered. The test may concern one component, a simulated workload or an earlier prototype. Begin by identifying the item, version, conditions and result. Then compare that scope with the product and use being purchased. An excellent measurement remains limited evidence if the delivered configuration, environment or acceptance criteria differ materially from those in the report.

Testing provides observations, including uncertainty and limitations when properly documented. A useful report states how the sample was selected, how many measurements were made and what was excluded. A successful run does not establish the frequency of failure across repeated use. The NIST BenchQC publication, for example, specifies simulator-based benchmarking work. Its scope matters precisely because a credible research source should not be treated as evidence that an unrelated physical device has passed a customer’s acceptance test.

Validation and formal certification

Validation connects measurements to a purpose. A control component could meet its laboratory specification while producing unacceptable crosstalk after installation. A processor could run a demonstration while failing the buyer’s turnaround or accuracy requirement. Agree what evidence would establish suitability before commissioning the campaign. The result may validate a bounded research use without supporting continuous commercial operation. That is a meaningful finding when the boundary is explicit, rather than a failure to produce a universal endorsement.

Certification is a separate conformity-assessment activity. ISO’s description of conformity assessment distinguishes testing, inspection and certification, and explains that certification is performed by external bodies rather than ISO itself. A buyer should ask which scheme and requirements apply, who issued the statement and precisely what it covers. Verify validity, exclusions and any continuing conditions. A laboratory’s institutional reputation or participation in a publicly funded project is not interchangeable with a product certificate recognised for the proposed procurement.

Standards, independence and reproducibility

Quantum infrastructure is developing measurement capability alongside standards. The Qu-Test periodic reporting published through CORDIS describes industrial testing and work on harmonising terminology and methods. That is evidence of a programme and its reported activities. It does not certify every participant’s products. Likewise, the CEN-CENELEC quantum programme identifies standards work across the field; participation in a committee cannot establish that a product conforms to a completed requirement. Buyers need the actual document and its applicable status.

Independence has several dimensions. A third party can perform a vendor-defined test on a sample selected by the vendor. That may be useful, but differs from independent sampling against buyer-defined criteria. Establish who paid, who set the method, who retained the raw records and whether adverse results could be withheld. Confidentiality need not prevent a sufficiently detailed report, but broad restrictions can make a public success statement impossible to assess. The buyer should know what remains unobservable.

Reproducibility determines whether evidence can travel. A result from one facility is more useful when another competent party can interpret the protocol and obtain materially consistent observations within the stated uncertainty. Device revisions, software changes and recalibration can alter that relationship. Record the method version and the changes that trigger retesting. For a supplier making repeated deliveries, evidence on variation across units may matter more than repeatedly demonstrating the strongest specimen.

The procurement decision should specify which report answers which requirement and which risks remain for integration or operation. A buyer can accept an independently tested component without claiming that an entire quantum system is certified. It can also require additional validation because its use differs from the test environment. Clear boundaries reduce duplicate testing where evidence is portable and preserve necessary tests where it is not. That is the practical contribution of credible measurement to a commercial transaction.