Why 300 KB is hard
An uncompressed A4 page scanned in colour at 300 dpi is about 25 MB of raw pixel data. Even after normal JPG compression inside a PDF, a single colour page from a scanner or phone is often 500 KB to 2 MB. So a 300 KB limit is not asking you to trim a little — it is asking you to remove 80 to 95 percent of the data. Where you remove it decides whether the result is readable.
Three things determine the final size of a scanned page: the number of pixels (resolution), how many colours each pixel can have, and how aggressively the image is compressed. Text needs enough pixels to keep letter shapes distinct. It does not need colour, and it tolerates moderate JPG compression. That is the whole strategy: spend the budget on pixels, not on colour or on pages nobody asked for.
Step 1: Cut pages before you compress anything
Size is shared between pages. With a 300 KB limit, a two-page document gets 150 KB per page; a ten-page document gets 30 KB per page. The first question is always whether every page needs to be there.
- Remove blank backs of single-sided documents that went through a double-sided scanner.
- Remove cover sheets, fax headers and "this page intentionally left blank" pages.
- If the form asks for "proof of address", it usually needs the page with your name and address, not all six pages of the bank statement. Read what is actually asked.
Rotate & reorder lets you delete pages visually, and Split extracts only the pages you name. Both run in your browser and do not reduce quality.
Step 2: Get rid of colour where it adds nothing
A colour scan stores three channels of information for every pixel. For a black-and-white document, two of them are mostly noise: paper tint, shadows, the slight yellow of office lighting. If you are scanning again anyway, choose greyscale or black-and-white in the scanner settings. A greyscale scan of a text page is typically a third to a half of the size of the colour version, before any other compression.
Keep colour only when it carries meaning: a coloured stamp, a photo on an ID card, a signature in blue ink that the receiving office wants to see as original.
Step 3: Choose the right resolution for the budget
For reading on screen, about 100 dpi is the practical lower bound for normal 10–12 point text. At that resolution an A4 page is roughly 830 × 1170 pixels. Footnotes and small print need more — around 150 dpi. Below about 75 dpi, lowercase letters like e, c and o start to blend together.
A useful rough guide for a greyscale text page compressed as JPG at moderate quality:
- around 150 dpi: often 80–150 KB per page
- around 100 dpi: often 40–80 KB per page
- around 75 dpi: often 25–50 KB per page
These numbers vary a lot with content — a dense spreadsheet is heavier than a letter with wide margins — but they show why a three-page document fits comfortably in 300 KB and a twelve-page one struggles.
Step 4: Let a tool search for the size
Trying settings by hand is slow. Compress to target size does the search for you: you type the limit and it finds the highest resolution that fits. Internally it starts at JPG quality 75 and looks for the best resolution between 30 and 200 dpi, and only lowers JPG quality if even 30 dpi is too big. That order is deliberate: dropping resolution a little costs less legibility than heavy JPG artifacts around every letter.
Type a number slightly below the real limit — 290 KB for a 300 KB portal. Websites disagree about whether a kilobyte is 1,000 or 1,024 bytes, and a file of 301 KB by one count is 294 KB by the other.
If you prefer to keep everything on your own computer, the browser-based Compress tool does the same kind of re-encoding with fixed levels. Start at High (1024 px wide, quality 60), check the size, and adjust the max width if needed.
Step 5: Check the result like the reader will
Open the compressed file and zoom to 100 percent, then to 150 percent, on the page with the smallest text. Check the things that matter to the receiver:
- Names, dates of birth and document numbers — every digit must be unambiguous.
- Stamps and seals — often lighter than the text around them and the first thing to disappear.
- Signatures — should still look like a signature, not a smudge.
If any of these fail, go back to step 1 and cut another page, rather than compressing harder.
When the document is not a scan
If your PDF was exported from Word or another program, its text is stored as text, not as pixels. Text is tiny: a ten-page report with no images might be 100 KB. When such a file is large, the weight is in embedded images — a logo at 4000 pixels wide, a photo pasted in at camera resolution. In that case, image-based compression throws away the efficient text layer and can even make the file larger. Better options are:
- Resize the images in the source document and export again.
- Use high-quality compression, which resamples embedded images to about 150 dpi but keeps text as text.
If it still will not fit
Some combinations are not possible: fifteen pages of dense small print will not be readable in 300 KB. At that point the honest options are to check whether the portal accepts several files, to upload only the pages required, or to contact the receiving office. Most offices would rather get a legible partial document with an explanation than an unreadable complete one.
Quick checklist
- Remove every page that was not asked for.
- Rescan in greyscale if colour is not needed.
- Target around 100–150 dpi for text.
- Aim 10 KB under the limit.
- Check names, numbers, stamps and signatures at 100 % zoom.