When OCR’ing a document in a smart rule, the OCR process runs in the background (I assume). So, a subsequent action that relies on PDF having a text layer (like “Scan text” or “text contains”) might fail because the OCR process has not terminated when the action runs.
The scripting suite’s OCR command provides a waitingForReply parameter. I suggest adding something similar to the smart rule action. Eventually 
That’s the job of the OCR & Continue action.
Of course you had thought of it. Thanks.
P.S: In both cases the smart rule actually waits for the completion of the OCR process. The only difference is that following actions either still use the original item (OCR action) or the created item (OCR & Continue). The same applies to e.g. Duplicate & Continue or Convert & Continue
So, when I set “move original to trash” in my OCR preferences, the following actions are performed on the document in the trash? That does explain what I saw (empty pages were not removed after OCR by my script).
This setting isn’t used by smart rules, it’s up to the smart rule to trash items or to use the right OCR action.
I don’t see any differences:
I must admit that I OCRd PDFs that already had a text layer, but that shouldn’t make a difference, or should it?
A screenshot of the smart rule would of course be useful.
OCR > Apply doesn’t create a new document, the existing one is updated.
