Why scanner names mislead by design
img20240315_0001.pdf contains a date, but the wrong one: it is the day the page was scanned, not the day the document was issued. For mail that sat in the inbox tray for two weeks, ten days easily separate the two — enough to misplace a deadline. The counter runs per stack and restarts at one the next morning, so filenames from different days collide in a shared folder. And the most important part is missing entirely: who wrote it and what it is about. Both of those sit inside the document itself.
Text layer or pure image — the decisive difference
A scanned PDF is at first only a photograph of paper. Some scanners place a text layer underneath when saving, many do not, or do it only for the first page. Filery first checks whether text exists inside the PDF and falls back to text recognition only when none is there. OCR is optional to switch on, because it costs time and is unnecessary for cleanly generated digital PDFs. You can tell whether a file has a text layer yourself, by the way: if text can be selected in a PDF viewer, it is there.
Why img20240315_0001.pdf becomes 2024-03-15_Stadtwerke_Abrechnung.pdf
In this example the scan date happens to coincide with the document date — for freshly opened mail that is the normal case. Filery nonetheless does not take the date from the old filename but the one from the letterhead, because only that one is reliable. Stadtwerke comes from the sender field, Abrechnung from the subject line. Following the same pattern, a bank statement, an official notice and a tradesman invoice in the same stack each get their own immediately readable name, without you having to sort the stack beforehand. You define the pattern once; field order and date format are freely adjustable.
When the date is missing or the stack is mixed
Three cases regularly cause trouble with scans. First, documents without a printed date, such as forms or cover letters carrying only a postmark — there the date field stays empty instead of the scan day being guessed into it. Second, letters with two senders on one sheet, typically a broker cover note above the statement of the actual provider; the header at the top is not always the one you want to file under. Third, the combined scan, where several documents sit inside one PDF file. Filery renames files, not pages — split such bundles beforehand.
Scan settings that noticeably improve recognition
The hit rate depends less on the software than on the paper path in front of it. What works well is 300 dpi in greyscale; higher resolutions barely help with normal office print and only lengthen processing. Remove staples before feeding so pages do not run through skewed, and leave the scanner deskew function switched on. The critical area is the top margin: that is where sender and date sit, and exactly where creases and punch holes are. Password-protected PDFs cannot be read and have to be unlocked first. A saved scan profile removes the need to set this up each time.
What happens to the scanned content
The text is extracted from the PDF on your device, for image scans through the text recognition you switched on. That content goes encrypted into the AI analysis; only the recognised metadata comes back. The app then carries out the renaming on the device again, and the scans themselves are not stored permanently. Anyone who wants to run the analysis through their own keys uses the Custom plan at 5 euros a month with their own API keys. For trying it out, the free Starter plan with 50 documents a month is enough and needs no sign-up.
FAQ
Related topics
Download Now
Choose your platform and start with automatic document renaming
Windows
Windows 10 & 11
We are currently working on official code signing for Windows. The download is already available – Windows Defender or SmartScreen may show a warning because the app is not yet certified. Choose “More info” → “Run anyway” to install.
System requirements: macOS 10.14+, Windows 10/11
All downloads are transmitted securely via HTTPS.