Yep, exactly, when layer data survives the PDF export, it’s a huge help. We use it as a weak signal for clustering and object grouping, but never rely on it fully since it’s often inconsistent or stripped. When it’s there, accuracy and speed both improve noticeably.