How to Compress a PDF Under 2 MB (Without Destroying It)

Every portal with an upload limit has taught someone the same lesson the hard way: you scanned a 30-page contract, the file is 48 MB, and the form wants “maximum 2 MB”. Before you start randomly clicking compress buttons, it helps to understand why PDFs get huge — because the right fix depends entirely on the cause.

Why your PDF is so big

A PDF is a container, not an image. What’s inside decides the size:

1. Scanned pages stored as full-resolution images. This is the big one. A flatbed scanner defaulting to 300 DPI color produces a ~2–4 MB image per page. Thirty pages → 100 MB territory. No amount of clever compression fixes a 300 DPI color scan except making those images smaller.
2. Photos embedded at camera resolution. A 12-megapixel phone photo pasted into a slide or report ships at 3–5 MB inside the PDF, even if it displays at a quarter of the page.
3. Fonts, embedded fully. Complete CJK font subsets can add tens of megabytes in pathological cases; ordinary Latin fonts add a few hundred KB per family.
4. Redundant structure. Editing history, unused objects, and old-form revisions accumulate in files edited over years. Saving clean (with a tool that rebuilds the document) can shave real percentages.
5. Vector content, rarely. True vector PDFs (text + drawings) are usually tiny — if a text-only PDF exceeds a few MB, it’s almost certainly carrying embedded images.

Quick diagnostic: if your PDF’s text is selectable and sharp when you zoom, and it’s still huge — it’s carrying heavy images or fonts. If the text is fuzzy when you zoom in, the whole document is already one big image (a scan), and you’re in case 1.

Match the method to the document

Method 1: Downsample and re-encode (the heavy hammer)

The brute-force approach renders every page to an image at a chosen quality and rebuilds the PDF from those images. Size drops dramatically because it attacks the dominant cost directly.

The honest trade-off: text stops being selectable and stops being text. Search inside the document dies (unless the tool adds an OCR layer), zooming shows raster fuzz, and screen readers lose the content. For a scanned contract that was never text anyway — the scan was an image — this costs you nothing. For a text-heavy report that someone will need to copy from, it’s the wrong tool.

Our compress PDF tool works this way, with a quality slider: you trade visible JPEG quality against final size, and you see the result before committing. It runs in your browser via WebAssembly — the document is rendered and rebuilt locally, never uploaded. For scanned documents it’s exactly the right hammer; run a page or two first, check the smallest text is still legible at 100% zoom, then commit the whole file.

Method 2: Split it

If the requirement is “under 2 MB per upload” (many court-filing and visa portals are), splitting a 40-page scan into two 20-page files may be simpler than degrading every page. Split PDF carves a document by page ranges in the browser; combined with moderate compression this is the standard workflow for portal limits. If you only actually need certain pages — the signature pages of a 60-page agreement — extract PDF pages instead of sending the whole thing at all.

Method 3: Fix it at the source (the best one)

The smallest possible PDF is the one that never carries dead weight:

• Re-export, don’t re-save. In Word, Google Docs, or Pages, use File → Export/Download as PDF rather than printing-to-PDF repeatedly. Fresh exports rebuild images at sane resolutions.
• Compress images before inserting them. If a photo will print at 8 cm wide, it doesn’t need 12 megapixels. Resize it first; the PDF inherits the savings.
• Scan sanely. 300 DPI, grayscale (not color) for text documents, JPEG quality 80–85. A black-and-white text contract scanned this way runs ~150–400 KB per page instead of 3 MB. Scanning receipts at 600 DPI color is a museum-archiving decision, not an office one.
• Strip what you don’t need. Metadata, form revisions and embedded thumbnails add up on old files. Our remove PDF metadata tool rebuilds the document without the private extras — which is also the privacy-conscious move before you email anything.

How small can you realistically get?

Rough field numbers for a 20-page scanned text document:

Treatment	Typical result
Original scan (300 DPI, color)	40–70 MB
Grayscale re-scan or conversion	8–15 MB
Rasterized compress, quality 60–70	3–8 MB
Rasterized compress, quality 50 + grayscale	1.5–4 MB
Same, plus split into two files	~0.8–2 MB each

Getting 48 MB under 2 MB usually takes two moves, not one: grayscale/downsample at the source or a firm compression pass, then split if the last megabytes resist. Push the quality slider only as low as the smallest text in the document stays readable — that line, not the overall look, is where a PDF stops being acceptable.

FAQ

Will compressing a PDF reduce its quality?

Rasterizing compression always costs something — the question is whether you can see it. At quality settings that keep the smallest text legible, most documents look fine on screen and print acceptably. Text-only PDFs (no scans) should never be rasterized; fix those at the source instead.

Why is my 2-page PDF 30 MB?

Almost always one or two embedded photos at full camera resolution. Extract the images, resize them to their display size, re-insert, re-export. Or run a compression pass — rendering those pages at display resolution does the same job after the fact.

Is it safe to upload a PDF to an online compressor?

It depends entirely on the service’s handling of your file. Contract-heavy and personal documents deserve caution: many services store uploads, some for weeks. Compression can also be done locally — our tools run entirely in your browser, and you can verify that with the network tab open while converting. For tax documents and contracts, local is the right default.

Does compressing a PDF remove its text?

Only if the method rasterizes pages (the compress tool does — by design, since it’s the only reliable way to shrink scans). Splitting, extracting pages and metadata removal all keep text intact. If you need searchable text and a small file, compress a copy and keep the original for search and copy-paste.