What the automated pre-checks look at
Every time you add a document or certificate, import data from a CSV file or send it through the API, DPPC runs a set of rules that check data quality. The result is a list of specific warnings and suggestions, such as a missing certificate number, an expiry date earlier than the issue date or a file whose name does not match the product. Errors show up right away instead of during review or after the product passport is published.
Certificates: data, dates and duplicates
For a certificate, the system checks whether the issuer and number are filled in, whether the expiry date comes after the issue date, whether the certificate is already expired when saved or expires soon, and whether its status matches its dates. It also detects a certificate with the same name saved earlier under a different number or issuer. It warns when the PDF file name does not match the certificate name or number, or when the file is suspiciously small and may be incomplete.
Product documents: does the file match the record?
For product documents, DPPC flags a document with the same title and type added a second time, very short titles and files whose names contain neither the SKU nor the product name. If the PDF has a readable text layer, the system also checks whether the document mentions the product and its own title, and whether it contains a different SKU, GTIN, batch number or serial number than the product record. It is a simple way to catch a document attached to the wrong product.
Data imports: products, materials and suppliers
The same rules apply to CSV imports. For products, the system flags a missing brand, model or country of origin, a description that is too generic, an unusual GTIN format and empty industry fields. For materials, it detects shares that add up to more than 100 percent, suspiciously low values, mass without a unit, a missing supplier and a country of origin that differs from the supplier country. For suppliers, it points out missing contact details or country and a VAT ID already assigned to another supplier.
Warnings, suggestions and a human decision
The checks never reject anything on their own. A document or certificate with warnings goes to review with the status "needs review", and the person in charge sees what each warning is about. If the readable PDF text contains a certificate number or issuer and that field in the record is empty, the system can suggest it. Every result also includes a confidence score. Through the API, the same checks are available in a preview mode that returns warnings without saving any data.
What the automated checks do not do
We are upfront about this, because trust in the process depends on it. The checks are predictable, transparent rules, not an artificial intelligence model. They do not recognize text in scans, and they only read content from PDF files that have a plain text layer. They also do not judge whether a document is correct in substance. Their job is to catch gaps, inconsistencies and mistakes before a person looks at the document, and the approval decision is always made by the person reviewing it.