the system does not understand a question from the list, what happens then

Unrecognized question gets a blank line, not a guessed answer

If the system does not recognize a question, that question remains open. There is no filled-in answer that "is probably good enough" — a question is only suggested if there is a fact in the chain file that matches it directly. If the system does not recognize the question, you will see that explicitly: blank, with the option to fill it in yourself or skip it. You must enter that skipped question yourself in your customer's portal, just like any other answer.

For a question that is worded slightly differently from what is already in the file — for example, a question about the ten measures from Article 21(2) of the NIS2 Directive, while the file is structured around the control measures in Annex A of ISO/IEC 27001 — it may happen that the system does not make the connection. The same applies to lists that combine questions, such as a question that wants both the number of employees and the locations of establishments in one field. The system often does not recognize such combined questions as a whole, and then manual entry is the quickest way.

Why no answer appears without a source

This follows from how the platform is set up: a "yes" is never suggested without an underlying fact in the file to support it. For some questions, simply no source data exists yet — for example, for a newly added question about an AI system, while whether the undertaking deploys AI systems has not yet been recorded as a separate piece of data. The same also applies to questions that refer to a voluntary standard such as the EFRAG VSME Basic Module, or to the reporting obligations under Directive (EU) 2022/2464: without a recorded answer in the file, the field remains blank. This is not a malfunction, but the absence of a match between question wording and file field.