Updated: August 1, 2026
Quick answer: PDF size usually comes from images, scans, embedded fonts, and duplicated objects. Start with balanced compression, check text selection and links, and use local processing for sensitive documents instead of uploading them to an unknown service.
“Compress a PDF” can mean several different operations. A text-native report may need only object cleanup and image recompression. A scanned contract may be almost entirely page images. A portfolio may contain large photos, fonts, thumbnails, and embedded media. Choosing the strongest preset without identifying the document type is how text becomes fuzzy, forms stop working, or a supposedly small PDF becomes unusable when printed.
The safe workflow is: make a copy, inspect what makes it large, choose a quality target, compress, and test the output in the way the recipient will use it.
First classify the PDF
Open the file and answer four questions:
- Can you select and search the text, or is every page one flat image?
- Are the pages mostly text, photographs, diagrams, or screenshots?
- Does the recipient need to print it, zoom into drawings, fill forms, or follow links?
- Does the file contain confidential content that must not reach a third-party server?
| PDF type | Usual size source | Main risk during compression |
|---|---|---|
| Text-native report | Embedded images, fonts, attachments, duplicate objects | Broken links, forms, or fonts |
| Scanned document | Full-page raster images | Unreadable small text and lost OCR |
| Photo portfolio | High-resolution color images | Soft detail and banding |
| Slide export | Screenshots, large backgrounds, embedded fonts | Small labels become fuzzy |
| Form or signed document | Images plus interactive/signature objects | Invalid signature or broken fields |
The best compression setting is the one that solves the actual bottleneck while keeping the required behavior.
Where PDF bytes usually come from
- Oversized images: A page may display an image at a few hundred pixels wide while the PDF embeds a camera original.
- Scanned pages: A scan stores the entire page as pixels. Reducing image resolution and color depth is often the largest lever.
- Embedded fonts: Fonts help the document render consistently, but a file may contain unnecessary full font programs instead of subsets.
- Duplicate resources: Repeated images, thumbnails, unused form objects, attachments, and editing remnants add bytes.
- Print-oriented exports: A screen-only document exported at print resolution carries detail the reader cannot use.
Text and vector shapes generally scale well because they are represented as structured data. If a text-heavy PDF remains huge, look for embedded images, attachments, or multiple font families before lowering every quality setting.
How PDF compression changes the file
Most optimizers combine several techniques:
| Technique | Typical effect | What to check |
|---|---|---|
| Downsampling | Fewer pixels in embedded images | Zoom and print sharpness |
| Image recompression | Fewer bytes per image | Photos, screenshots, and fine text |
| Grayscale conversion | Removes color channels | Whether color carries meaning |
| Font subsetting | Keeps only used glyphs | Font appearance and language coverage |
| Object cleanup | Removes unused or duplicated data | Links, forms, bookmarks, attachments |
| Linearization | Reorders data for progressive web viewing | Browser loading, not usually total bytes |
| OCR or text-layer changes | Adds or alters searchable text | Selection, search, copy, and accessibility |
Adobe's PDF optimization guidance describes these categories in the context of Acrobat. Other tools expose different names, but the trade-offs are similar.
A practical online workflow
Use an online service only for documents that are appropriate to upload. Before choosing one, check the current privacy statement, retention period, file-size limit, and whether the free tier changes the output.
- Save the original. Work from a copy and record its size and page count.
- Choose the use case. Screen reading, email attachment, print, archive, and form submission need different settings.
- Run a balanced or standard pass. Avoid the strongest preset as the first attempt.
- Download with a new filename. Keep the original separate from the delivery copy.
- Inspect every important feature. Search, select, zoom, links, bookmarks, forms, signatures, images, and page order.
- Check the byte size and destination. A file that passes a form upload limit but fails to open on a phone is not a successful result.
For a routine web handout, a balanced pass is usually a better first experiment than forcing a specific target such as 1 MB. If a target is mandatory, make the trade-off explicit and test readability after every stronger pass.
Choosing resolution and image quality
The right image resolution is tied to the final use:
- Screen-only reading: A lower image resolution can be sufficient, especially when the page is not expected to be zoomed heavily.
- Office printing: Keep enough resolution for small type, line art, and diagrams; compare a printed page rather than judging only on screen.
- Photo-heavy documents: Use a quality setting that preserves faces, edges, and gradients; converting everything to aggressive grayscale may save bytes but change the document's purpose.
- Scanned text: Optimize the scan, then inspect characters at 100% zoom. A scan that looks fine as a thumbnail may fail when the recipient searches or prints it.
Treat common DPI values as starting points, not guarantees. A 72 DPI export may be acceptable for a small screen preview and unacceptable for a document with fine print. The image's displayed size and the smallest meaningful detail matter more than a single preset number.
Getting a PDF under a hard limit
When an upload form imposes a limit, use a measurement loop:
| Pass | Change | Stop when |
|---|---|---|
| 1 | Remove redundant objects and recompress oversized images | The file is smaller with no visible change |
| 2 | Resize images to their actual display size | Fine text and diagrams remain clear |
| 3 | Convert unnecessary color scans to grayscale | Color is not carrying information |
| 4 | Reduce photo quality or scan resolution | The output still passes the recipient's readability test |
| 5 | Rebuild or split the document only if the workflow allows it | The target is met without losing required pages or features |
A scanned color book may simply not fit a very small limit without losing useful detail. In that case, ask whether the recipient accepts a ZIP, a split document, or a lower-resolution reading copy, and retain a high-quality master.
When online compression is the wrong choice
Use local processing for passports, IDs, medical records, financial statements, signed contracts, private client work, and unreleased business material. HTTPS protects a connection; it does not turn an upload into local processing or tell you how long a provider retains the file.
Local options include a desktop PDF optimizer, Acrobat's optimization panel, Preview for basic exports, PDF24 Desktop, or a carefully reviewed Ghostscript workflow. The interface matters less than keeping the source on your device and validating the output.
Do not use “Print to PDF” as a universal compression fix. It may flatten forms, remove links and bookmarks, alter accessibility, invalidate signatures, or rasterize selectable text. Use it only when those changes are acceptable and the resulting PDF has been tested.
Post-compression quality checklist
- Page count and page order are unchanged.
- Text can still be selected, searched, and copied where expected.
- Small labels, charts, and scanned characters remain readable.
- Links, bookmarks, forms, and attachments still work.
- Digital signatures are still valid, or you intentionally created a new unsigned copy.
- Color is preserved where it carries meaning.
- The PDF opens in a browser, desktop viewer, and mobile viewer if those are in scope.
- Print a representative page if printing matters.
- The original high-quality file is archived separately.
FAQ
Can I compress a PDF online for free without installing anything?
Yes, for a non-sensitive file. A browser service can downsample images and clean objects without a desktop installation. Check current file limits, retention terms, and output behavior before uploading.
Will compressing a PDF reduce its quality?
It can. Object cleanup and font subsetting may be visually lossless, while image downsampling, recompression, and grayscale conversion change the rendered result. Use the lightest setting that meets the size requirement.
How do I compress a PDF below 1 MB?
Identify the largest component first, then resize images to the actual use, remove unnecessary color, and use a stronger image setting only after checking readability. Some image-heavy documents cannot reach 1 MB without a meaningful quality trade-off.
Is it safe to upload a sensitive PDF to an online compressor?
Not by default. Keep confidential documents local unless you have evaluated the provider, its terms, and your organization's data policy. A local workflow is easier to control.
Why is my PDF still large after compression?
High-resolution scans, embedded attachments, duplicate images, and full fonts are common causes. Inspect the structure and target the dominant object rather than applying a stronger preset to everything.
How do I compress a PDF offline?
Use a local optimizer, desktop application, or trusted command-line workflow. Make a copy and verify text, links, forms, signatures, print output, and page order after the export.
Sources
Next steps
Take one representative PDF page and compare a balanced export with a stronger export at 100% zoom and on paper if relevant. Choose the smallest version that preserves the features your recipient actually needs, then use that setting for similar documents.