Every recognition or data extraction activity consumes page units.
The A4 format is the base unit (which is also the smallest unit); its area ratio is decisive, with a +5% tolerance.
Page = fuzzy A4 area (+5% tolerance; the smallest unit is A4)
We count a page unit only once per page if it undergoes any recognition process for data extraction purposes (OCR, ICR, OMR, etc.) and/or a data extraction process, and even then, only once per page.
When creating a PDF with text layers during export, we do not charge page units for OCR (since the OCR process is not intended for data extraction).
If the PDF does not specify a resolution, it will be converted to 300 dpi, and we calculate its size relative to the page unit (compared to A4) based on that.