Firms spend enormous effort on the narrative sections of regulatory submissions — the parts explaining context, rationale, and intent. Supervisors, in my experience, read those sections quickly and spend their real attention on the numbers and the annexes, checking whether the submission's own data supports the narrative it's telling.
This is a predictable consequence of how supervisory review actually works: a reviewer with limited time and a stack of submissions is looking for inconsistency, because inconsistency is the fastest signal that a submission's narrative doesn't match its underlying reality. A beautifully written explanation sitting next to a data table that doesn't quite support it is a red flag, not a reassurance.
Firms that treat regulatory writing as a persuasion exercise are optimising for the wrong reader. The submissions that fare best are the ones where the narrative is almost unnecessary — where the data tells the story clearly enough that the accompanying text is simply confirming what a careful reader would already conclude.
This has a practical implication for how submissions get built internally: the data and the narrative should be produced by people who can see both simultaneously, checking for consistency as they go, rather than a data team and a communications team working separately and reconciling at the end. Inconsistencies introduced late are the ones most likely to survive into the final submission.
The firms with the smoothest supervisory relationships aren't the ones with the best writers. They're the ones whose numbers and narrative were never allowed to drift apart in the first place.