1. Check what the limit number actually means
A "1 MB" limit can mean 1,000,000 bytes or 1,048,576 bytes (1,024²), depending on which convention the portal's code uses — and file managers on different operating systems don't always agree either. This isn't a size-reduction method, it's the first thing to get right: if your file is reported as 1.01 MB and the limit is "1 MB," you may only need a tiny trim, not a full recompression pass. When in doubt, aim comfortably under the stated number rather than right at it.
2. Remove pages that weren't asked for
This is usually the single biggest win, because the limit is shared across every page. Blank backs from double-sided scans, cover sheets, and pages the form doesn't actually require are free size reductions with zero quality cost. Rotate & reorder deletes pages visually; Split extracts only the pages you name.
3. Lower the resolution of scanned pages
If the file is a scan or a photographed document, resolution is usually where most of the size lives. Dropping from, say, 300 dpi to 150 dpi for a scanned page cuts the pixel count to roughly a quarter, which translates to a large drop in file size for a modest drop in sharpness — text generally stays readable down to around 100–150 dpi. Compress to target size automates this search for you.
4. Strip colour where it isn't needed
A greyscale version of a colour scan is typically a third to half the size, before any other compression. Worth doing at the source (rescanning in greyscale) rather than converting afterward, if the original document is still available to you.
5. Fonts and metadata — usually a minor gain, occasionally a real one
Metadata (author, creation date, thumbnail previews) adds only a few kilobytes on most files — not a serious lever for a size problem by itself, though Remove metadata and Edit metadata are worth using for privacy regardless. Embedded fonts are a different story: a PDF exported from Word or similar software can embed an entire font family — sometimes several hundred kilobytes — even if the document only uses a handful of characters from it. If your document software offers a "subset fonts" or "minimize file size" export option, use it; that can matter more than any compression pass applied afterward.
6. Split instead of compress, if the portal allows it
If the upload field accepts multiple files, splitting a long document into parts that each clear the limit avoids the quality trade-off entirely. Split extracts page ranges into separate files. This only works if the form genuinely accepts more than one upload — check first.
7. Let a target-size tool do the search
Once pages are cut and colour/resolution decisions are made, Compress to target size finds the highest resolution that fits a KB or MB number you type in, rather than you guessing at settings by hand. It now runs directly in your browser with no daily limit (a server-side mode is still available if you'd rather use it for a particular file). Compress does a similar re-encoding using fixed Low/Medium/High/Extreme levels instead of an exact target, if you don't need a precise figure. High-quality compression is the better choice instead, specifically for documents where the text itself must stay selectable and sharp — it resamples embedded images to about 150 dpi using Ghostscript's "ebook" preset but leaves text and vector drawings untouched.
A worked example
A five-page scanned form, 3.2 MB, needs to get under 500 KB. In order: check the limit means 500 × 1,024 bytes, not 500,000 (step 1) — a small but real difference. Two of the five pages are blank backs; removing them with Rotate & reorder drops the file to roughly 2 MB (step 2). The scan was done in colour but the content is black text on white paper, so rescanning in greyscale — or, if that's not possible, accepting the colour-to-greyscale conversion inside a compression pass — removes another third to half (step 4). What's left, three pages of now-greyscale scan, goes through Compress to target size set to 490 KB, which searches resolutions down to whatever fits (step 7). Metadata and fonts (step 5) rarely matter for a scanned document, since there's no embedded font to begin with — that lever matters more for files exported from word processors.
Typical impact of each method
| Method | Typical effect |
|---|---|
| Cutting unnecessary pages | Often the largest single reduction — proportional to pages removed |
| Greyscale instead of colour | Roughly a third to a half smaller |
| Lower resolution (e.g. 300→150 dpi) | Large reduction, scales roughly with the square of the pixel drop |
| Removing metadata | A few KB — minor by itself |
| Removing/subsetting embedded fonts | Negligible to several hundred KB, depending on the source document |
| Splitting into multiple files | No quality loss, but only works if multiple uploads are accepted |
Quick checklist
- Confirm what the limit number actually means before doing anything else.
- Cut pages first — it costs nothing in quality.
- Rescan or convert to greyscale if colour isn't needed.
- Use Compress to target size to search for the right resolution automatically.
- If the document is digital (not scanned), try high-quality compression before an image-based tool.
- Split rather than degrade quality further, if the portal allows more than one file.