I have sat through audits where every part on the line was conforming and we still lost the weekend. A heat-treat certificate linked to the wrong lot. A PPAP revision that had moved while the dossier slept. The parts were good; the record was scrap. TÜV SÜD has now launched what it calls the first certification mark for a Digital Battery Passport. I cannot think of another conformity mark that sits on a data artefact instead of on the product it describes. The EU battery passport obligation is months away, and most plants will meet their own passport for the first time during the audit.
What the mark certifies – and what it quietly doesn't
Read the scope before you celebrate. The mark covers data completeness, integrity of the links between fields, and plausibility checks. It does not certify the cell chemistry. It does not certify the process capability behind that chemistry, and it does not certify that the carbon footprint is true – only that the figure is present, linked and plausible. Certification bodies certify the management of things, and data has now become the thing being managed. A plant that buys the mark and assumes the battery itself has been certified has read it backwards.
Two decades across automotive and aerospace drilled one rule into me under IATF 16949 and EN 9100: a part without history is scrap. In aerospace we did not debate this. We binned parts with perfect geometry because their pedigree had a hole in it. The battery passport inverts the logic – the history is the part. The auditor of 2027 scans the QR code before anyone opens the cargo doors, and what loads on the tablet is your product.
The failure modes of a record
The failure modes are unglamorous, which is exactly why they will hurt.
Most of them are copy-forward errors. A supplier declaration lifted from a PPAP that has since changed revision stays true of a part nobody ships any more. Carbon footprints get calculated against a supplier mix that shifted two substitutions ago. Lot-links break the day rework splits a batch in two and only one half keeps its pedigree. And unit conversions – grams against kilograms of CO2e, ampere-hours against watt-hours – pass every plausibility check while misstating the battery to a regulator. Four nonconformities, none of them against a physical part the auditor will ever touch.
Every link in a passport is a promise made by someone who has since changed the process.
I learned this the expensive way while building the QA/QC function for a 900-strong greenfield plant. The findings that cost us sleep were broken record links, never the parts. Part of the 70% defect-cost reduction we eventually delivered came from treating documents as product characteristics with their own reaction plans – a wrong revision on a control plan was escalated with the same discipline as a wrong dimension. Later I designed and built an orchestration platform that runs more than 60 AI models in parallel, and the lesson transferred without translation: pipelines fail at hand-offs, not components. A battery passport is forty supplier hand-offs in a trench coat.
Put the passport in the control plan
If the record is a product characteristic, it earns the full treatment. Passport data goes into the PFMEA with the criticality of a safety characteristic: severity for wrong data that reaches a regulator, occurrence for the hand-off that produced it. The hand-offs are easy to name – the supplier portal, the rework cell, the conversion step in the middleware – and they are where this fails.
Then run measurement-system analysis on retrieval. Not on the gauge. On the retrieval. Can three people on three shifts find the same record, open the same link, read the same value? If the answer varies by shift, the passport has a reproducibility problem no server upgrade will fix.
Records get sampled the way dimensions do. Pull a handful of passports every month and verify them field by field against source documents; a wrong field you find internally costs an hour, the same field found by a market-surveillance authority is a six-figure day in €.
And name the owner – in quality, not IT. IT owns uptime, and they should. Quality owns truth. Write the reaction plan for data escapes before you need one: containment means flagging the record and notifying downstream users, correction means an 8D on the hand-off that corrupted it. A part can be reworked. A record can only be re-issued.
A preview, not an exception
The industry conversation this month is autonomous agents on the factory floor – vendors demonstrating C-suite AI and coordinated machines. Fine. But the dull artefacts, the records nobody demoed, are what the next decade of audits will turn on. Digital product passports are coming for sector after sector under the EU's wider product rules, and each will be audited the way this one now is: record first, product second.
Key takeaways
- Give passport data a line in the PFMEA with the criticality of a safety characteristic – severity scored on wrong data reaching a regulator, occurrence on the hand-off that creates it.
- Treat retrieval as a measurement system: run attribute agreement analysis on whether different people on different shifts retrieve and read the same field the same way.
- Sample passports against source documents monthly, field by field, and escalate a wrong field exactly as you would a wrong dimension – containment, 8D, re-issue.
- Put the record owner in quality, not IT, and pre-write the reaction plan for data escapes; a record cannot be reworked, only re-issued.
Passport data is a product characteristic. It needs a PFMEA line, a control plan entry and a reaction plan, because the auditor of 2027 audits the record first and the cell second. Stop arguing database versus part – the record is the part now. Better to find a broken link in your own audit than to read about it in a notified body's finding.