PDFEnrich · Reproducible PDF attachment study
What a PDF attachment weighs after email encoding
A PDF's folder size is not its full email footprint. We measured seven public fixtures byte-for-byte before and after standard Base64 encoding so attachment planning can use reproducible numbers.
Reviewed July 29, 2026 by PDFEnrich Product Engineering.
Compress a PDF for emailMethodology
Each fixture was read as raw bytes and encoded with standard RFC 4648 Base64. We recorded the source byte count, encoded byte count, percentage increase, and parseable page count where available. The malformed fixture and password-protected fixture remain in the set because email transport encodes bytes whether or not a reader can open them without recovery information.
What the numbers mean
Base64 represents every three source bytes with four encoded characters, with small rounding effects for files whose byte count is not divisible by three. The measured fixtures therefore grew by about one-third before MIME boundaries, filenames, message headers, body content, signatures, or other attachments.
How to plan an attachment target
Do not aim at the exact message ceiling. Provider, plan, recipient, and organization policies vary, and some systems measure the entire encoded message. Leave margin, verify the current recipient rule, and use an approved sharing workflow when compression would make the PDF unreadable.
Limitations and reproducibility
Seven synthetic fixtures do not represent every PDF or every mail system. The study measures encoding overhead, not delivery success, spam filtering, provider limits, or compression quality. Download the CSV, JSON method record, and source fixtures to repeat or extend the calculation.
Method references
Measured raw and Base64-encoded fixture sizes
| Fixture | Raw bytes | Base64 bytes | Increase | Pages |
|---|---|---|---|---|
| Complex layout | 3,611 | 4,816 | 33.37% | 2 |
| Encrypted sample | 2,977 | 3,972 | 33.42% | Password required |
| Large progressive | 154,441 | 205,924 | 33.34% | 100 |
| Malformed recovery | 83 | 112 | 34.94% | Not parseable |
| Clean scan | 129,801 | 173,068 | 33.33% | 1 |
| Degraded scan | 147,071 | 196,096 | 33.33% | 1 |
| Simple searchable | 2,583 | 3,444 | 33.33% | 2 |