Why “online compressor” usually means upload
Most web compressors begin by transferring your document to a remote server. That can be convenient, but it also means the file crosses a network boundary before any size reduction happens. For a public brochure this may be acceptable. For a contract, medical record, bank statement or unreleased presentation, it is a decision worth making deliberately.
A local browser tool changes the architecture: the page downloads application code, then your browser reads and rewrites the PDF in memory on your device. The document itself is never posted to an upload endpoint.
What actually makes a PDF smaller
PDF size usually comes from embedded images, scanned pages, duplicated resources and occasionally unused objects. Text and vector graphics are already compact. The biggest safe win is therefore selective image recompression: resize oversized images, choose an appropriate JPEG quality and leave text untouched.
Lossless cleanup can remove redundant structure without changing pixels, but it cannot promise dramatic savings. A trustworthy tool should say when a file is already efficient instead of pretending every document can shrink by 90%.
Choose a target based on the destination
- Email and web: prioritize a small attachment and readable screen output.
- Balanced: preserve comfortable zoom quality while cutting presentation and scan weight.
- Print: use lossless cleanup when every pixel matters.
Bindery computes each candidate from the actual document before you commit, so the displayed estimate is the resulting byte size—not a generic marketing range.
Step by step
- Open Compress PDF.
- Drop the PDF. It stays in the browser tab.
- Wait for the real output estimates.
- Choose Email & web, Balanced or Print.
- Download the new file and open it at 100% and 200% zoom.
- Keep the original until you have checked the result.
How to verify that no upload occurred
Open your browser developer tools, select the Network panel and process a sample document. You should see application assets load, but no request containing the PDF bytes. Bindery also works after the app has loaded and the connection is unavailable, which is a useful architectural check.
When not to compress
Do not repeatedly recompress archival masters, print-ready artwork or files whose embedded images are already optimized. If a smaller output does not save meaningful space, keep the original. Compression is a delivery optimization, not a ritual.
Scans need a different decision
A scanned PDF is essentially a stack of photographs, so it can shrink dramatically—but the smallest setting can soften fine print. Start with Balanced, inspect signatures and small type, and move to Email & web only when the document is intended for screen reading. If the scan also needs searchable text, run OCR after compression and verify a few names and numbers manually.
For evidence, records or archival material, keep an untouched master and label the compressed copy clearly. A smaller delivery copy should never become the only surviving version by accident.
A useful quality checklist
- Open the first, middle and final pages.
- Inspect the smallest text at 200% zoom.
- Check photographs for block artifacts and gradients for banding.
- Confirm links, selectable text and page dimensions still behave as expected.
- Compare the output byte size with the original instead of trusting a percentage alone.
Compression is successful only when the recipient can use the result. The best setting is therefore the smallest file that still serves the document’s actual purpose.