Scan Documents to PDF Once: Keep an Archive Copy and an Upload Copy

Most people scan a document for a reason: a visa form wants the birth certificate, a landlord wants three payslips, a university wants the degree. So they scan it to suit that one request, often at a low resolution so the file fits the upload limit, and move on. A year later a different portal asks for the same certificate with a different limit, or a bank wants a clearer copy, and the low-resolution file is not good enough. The paper is in a drawer somewhere, or no longer in your hands at all.

The fix is to separate the two jobs. The first file is made for keeping: scanned at a generous resolution, straightened, made searchable and saved in a format meant to last. You make it once and never compress it. The second file is made for sending: a copy of the first, cut down to whatever the current portal or inbox accepts. Upload copies are disposable, because you can always make another one from the archive file in a minute. The cards below compare the two.

The extra effort is small. Scanning at 300 ppi instead of 150 takes the same number of passes through the scanner, and a phone photo takes the same tap either way. What costs time is everything after: rescanning when a second request arrives, chasing a document you already handed back, or explaining to an official why a stamp is unreadable. Doing the careful version once moves all of that work to a single afternoon, after which each new request is a two-minute job at your desk.

Resolution is the one setting you cannot fix later. A useful benchmark comes from the US National Archives, whose rules for federal agencies digitizing permanent paper records set a minimum of 300 pixels per inch for modern textual documents, in color or grayscale, and 400 ppi for photographs and papers with fine detail. Those rules are written for government records, not for your kitchen drawer, but the reasoning carries over: 300 ppi keeps small print, stamps and signatures readable when someone zooms in, and gives text recognition enough detail to work with.

On a flatbed scanner, choose 300 ppi, color or grayscale rather than black and white, and save as PDF or as TIFF or PNG images. Pure black-and-white mode makes small files but throws away faint stamps, pencil notes and colored seals for good. With a phone, use a scanning app or the camera, fill the frame with the page, hold the phone parallel to the paper and use even daylight without a shadow from your hand. If the app saves photos rather than a PDF, the Image to PDF tool turns them into one document, ordered by Date taken so the pages line up in the order you shot them.

A scan is a picture of a page. You can read it, but your computer cannot: searching for a policy number or a surname finds nothing. Optical character recognition fixes that by reading the letters on each page and storing them as an invisible layer behind the image. OCR PDF produces exactly this with Output left on Searchable PDF: the page looks unchanged, but you can search, select and copy text. Straighten pages and Auto rotate are on by default and correct the tilted and upside-down pages that both feeders and phones produce; Enhance contrast helps with faded photocopies and thermal receipts.

Searchability is what makes an archive useful years later. A folder of fifty image-only scans named scan001 to scan050 is almost as hard to use as the paper. A folder of searchable PDFs lets your computer's own search find the letter that mentions a reference number. Recognition is good but not perfect, so the archive copy still keeps the page image as the authority; the text layer is a finding aid. If you later need to edit a scanned letter rather than search it, the guide to converting PDFs to Word by document type explains the extra steps for scans.

An ordinary PDF can depend on things outside itself: fonts installed on the computer that made it, color settings of a particular device, scripts, links to other files. That is fine today and a risk in twenty years. PDF/A is the ISO profile of PDF built to remove those dependencies. The Library of Congress summary of the PDF/A family lists the main rules: all fonts embedded, color defined in a device-independent way, standard metadata included, and encryption, JavaScript, audio and video not allowed.

For scans, the choice is mostly between two levels. The PDF/A-2b level guarantees that the pages will look the same in future, while level u adds that all text has Unicode equivalents, so it can be searched and copied reliably. The PDF to PDF/A tool converts a finished searchable PDF to either; Make scans searchable can also add the text layer during the conversion, and Check only tells you whether an existing file already conforms. If you build the PDF from photos, PDF/A (archive) in the image converter creates PDF/A-2b directly.

When a request comes in, start from a duplicate of the archive file, never the original. Read the limit next to the upload field and make the copy to fit it. The US National Visa Center's scanning tips are a typical example: one document per file, all its pages together, no more than 2 MB, and a suggested resolution of 150 DPI. A 300 ppi archive scan holds four times the pixels of a 150 ppi page, so there is plenty of room to come down without losing legibility.

Compress PDF does this in one step. Choose Target size, type the limit in KB or MB, and image quality is lowered step by step until the copy fits, while the text layer from your OCR pass stays in place. For black-and-white paperwork, Convert to grayscale saves a lot on its own. Remove metadata is worth switching on for anything sent to strangers, since scanning software can record names and device details. If the portal wants only some pages, or one file per document from a multi-document scan, the guide to splitting a PDF for upload portals walks through it.

PDF/A is not needed for every file. It shines for documents you may have to produce years from now, such as deeds, pension statements, diplomas and court papers, and for anything you hand to an archive or an office that asks for it. For a phone bill or a delivery note, a searchable ordinary PDF is enough. Converting does make files somewhat larger, because fonts and color profiles now travel inside them, which is one more reason the archive copy and the upload copy should be two different files.

An archive you cannot find things in is just a second drawer. Give every archive file a name that sorts by date and says what it is, for example 2026-10-14 Birth certificate Jane Doe.pdf, with the date the document was issued rather than the day you scanned it. Keep a small set of top-level folders that match how you look things up, such as Identity, Education, Home, Money, Health and Work, and avoid deep nesting; two levels are usually enough. The searchable text layer covers the rest.

Keep upload copies out of the archive. A separate Outbox folder, or a folder per application, stops a compressed 180 KB copy from replacing the real scan by accident, which is the most common way an archive slowly loses quality. Back up the archive folder to a second place, such as an external drive or a cloud account, and check once a year that the files still open. Because PDF/A files carry their fonts and color inside, they are the least likely part of the folder to cause trouble.

Color costs file size, so it should earn its place. Scan in color anything where color carries meaning: certificates with colored seals or security printing, passports and ID cards, photographs, maps, medical images, documents with colored signatures or stamps, and anything an official might need to check as an original. The National Archives rules above allow color or grayscale for textual records, which tells you that plain grayscale is a respectable archive choice for ordinary paperwork.

Grayscale suits typed letters, statements, contracts and printed forms. It keeps faint pencil marks and the shading of stamps that black-and-white mode would destroy, at roughly a third of the data of color. For the upload copy, grayscale is often the single biggest saving, and many portals accept it for everything except identity documents and photos. When in doubt, scan the archive copy in color; you can always make a grayscale upload copy from it, but you can never add color back to a grayscale scan. Whatever you choose, use the same mode for every page of one document, so a reviewer never wonders whether a grayscale page is a later copy.