There are two broad ways for software to name a PDF from what is printed on it, and they suit different paper.
Whole-page reading. The tool reads all the text on the page and works out which part is the invoice number, the date, or the vendor. That is flexible, because it does not care where on the page a value sits, so it can cope with documents from many senders in many layouts. The cost is predictability. When the tool picks the wrong number, such as a purchase order number instead of the invoice number, nothing about the layout tells it so, and the mistake looks like any other name.
Zone reading. You tell the tool where to look. ScanRoute reads one zone you draw over a reference line, on the first content page of each document, with the Windows OCR language installed on your PC. A pattern splits that line into named fields, and each field is checked against what it should be: a field set to digits must contain only digits, a field set to letters only letters, and a date must parse as a date. Because the zone always reads the same spot, a wrong value tends to break the pattern or fail a check, and the document stops in the inbox instead of being filed under a wrong name.
Which paper suits a zone. Zone reading works when the reference line is printed in the same place every time: invoices from your own billing system, statements from one bank, forms your office prints, reports from one inspection template. It works poorly on a pile of letters from different senders. If your paper is mixed, run it in batches by layout, or accept that the odd ones will wait in the inbox for a typed value. ScanRoute reads one zone per document and does not read the whole page to choose a name.
Getting a clean read. Draw the zone around the reference line with a small margin and nothing else inside it. Choose printed text, not anything written in by pen. If no OCR language is installed, add one in Windows language settings; until then, splitting still works and the fields wait in the inbox for you to type.
From fields to a name. A template turns the fields into a path. {Client}\{Date}_{Type}_{Id}.pdf creates a folder per client and a sortable name inside it. Values that cannot go in a Windows file name, such as a slash, are flagged as bad file names rather than silently changed, and two documents that would get the same name are both flagged as duplicates rather than one overwriting the other.
Renaming means writing a named copy. ScanRoute does not rename the original file in place. It writes each document as a new PDF in the destination folder you choose, when you commit from Preview, and the original is only read. That is why it suits a combined scan: one source file becomes several named documents, and the scan itself stays exactly as the copier saved it. Each commit also leaves a CSV manifest listing every file written with its SHA-256 hash.