How small calibration labs can turn PDF certificates into DCCs

A PDF certificate needs a person to read it. NPL now sends machine-readable certificates for several of its thermometry services through an online portal it describes as “a demonstrator” (NPL, 2026). UKAS manages 366 calibration accreditations (UKAS, 2026), and the certificates behind them are the documents this shift will reach.

In UK calibration specifically, the person with most at stake is the technical leader of an accredited lab with 10 to 40 staff. Their certificates are assembled from instrument software, spreadsheets and a template, and leave the building as PDFs. In UKAS guidance, a data format is one option.

This article argues that readiness for digital calibration certificates (DCCs) is decided upstream of the certificate, in how results move through the lab. It shows where that flow breaks, then gives three actions and a five-question scorecard to use before customers ask. It belongs to our sector insight series.

DCC readiness is a data-flow issue

A lab’s readiness for DCCs depends on how cleanly its results travel from instrument to certificate. NPL defines a DCC as “a text file that provides all the information in a standardised form that a computer can read” (NPL). Software checks that file against published schemas, so it exposes every weak hop in the data behind it.

• Results live in several places. In NPL’s own case study, calibration data sat in two Excel workbooks whose cells were “typed in or pasted”, a step the authors call “subject to error” (NPL, 2024).
• Certificates are built for the eye. NPL’s minimal DCC template still needed at least 25 text elements, each mapped to a defined field in the record (NPL, 2024).
• Signing and delivery need new tools. NPL held digital signing and secure delivery back from its first case study because they required “a longer timescale” (NPL, 2024).

In summary, the format change itself is small, and the preparation it demands sits in the lab’s data flow.

Why does structured certificate data matter for your customers?

Structured data matters because a customer’s first job with any certificate is to move its numbers into their own quality system. NPL describes that transfer from a PDF as “time-consuming and, more significantly, prone to error” (NPL, 2024). With a DCC, the customer’s software reads the values directly.

The international bodies have chosen a direction. Since March 2022, 12 international organisations, including ISO and the Global Accreditation Cooperation, have signed up to “a world-wide uniform and secure data exchange format based on the International System of Units” (BIPM, 2022–2026). The Global Accreditation Cooperation replaced ILAC and IAF on 1 January 2026, so the body that coordinates laboratory accreditation worldwide sits among the signatories (IAF, 2026).

What goes wrong when a lab treats a DCC as an export format?

A lab that treats the DCC as a new export button meets its problem one step earlier. The PDF looks finished. The weakness sits in the copies made before it, as a value moves from reference software to a workbook, then into an uncertainty budget and finally into a certificate template.

We call this the retyping layer: every manual copy between a measurement and the certificate that reports it. An export tool placed on top of the retyping layer produces a machine-readable file carrying the same slips, and those slips become harder to spot once a customer’s software accepts them as data. When we map a certificate workflow, the retyping layer is the first thing we count.

Three actions for technical leaders to prepare a lab for DCCs

Mapping — tracing every hop from instrument to certificate

Start with one certificate type and draw the path a single reported value takes, from the reading on the reference instrument to the number on the signed certificate. Each hand-off where a person types or pastes is a hop in the retyping layer. A DCC built on that path inherits whatever slips occur at those hops, and the finished map shows which certificate types need the most work before structured output makes sense.

Governing — giving each certificate field one source

Once the hops are visible, assign every certificate field to a single owning system. The uncertainty value, for example, should come from one budget file that the certificate reads, so a correction made there reaches every output and lowers the risk of a reissued PDF. This is ordinary data governance.

Piloting — issuing one DCC alongside the PDF

Choose one instrument type and one customer whose quality team would use the data, and issue a DCC alongside the usual PDF. UKAS LAB 5 already allows results to be delivered in a data format agreed with the customer, so the pilot runs inside current accreditation guidance (UKAS, 2025). Keep the PDF as the human-readable copy until the customer confirms their system reads the file correctly.

How ready is your lab for digital calibration certificates?

Your score on five questions shows how much work sits ahead of you. Score each question 0, 1 or 2. Then read the band for your total.

Question	0	1	2
Can you name every point where a reported value is typed or pasted?	No	Partly	Yes, documented
Does each certificate field come from one owning system?	No	Partly	Yes, documented
Is each uncertainty stored as a number with its unit and coverage factor?	No	Partly	Yes, documented
Do you know which customers want calibration results as data?	No	Partly	Yes, documented
Is one named person responsible for DCC schema versions and signing?	No	Partly	Yes, documented

0–3: Map one certificate path this month before looking at software. 4–7: Fix field ownership for your highest-volume instrument type, then plan a pilot. 8–10: Offer a DCC pilot to one customer and ask NPL about its training and guidance for calibration labs.

What to do next

This method stops at the edge of the lab. Digital signing and the choice of schema version depend on bodies outside the lab: PTB maintains the DCC schema, and software suppliers decide when their tools support it. UK figures on DCC uptake are still unpublished, so the timing of customer demand remains a judgement.

Readiness is decided upstream of the certificate, in the retyping layer, and you can start measuring it this week. Take last week’s five certificates and count the manual copies behind each reported value.