By James 
Run a screenshot through a compressor and it drops 67%. Run an iPhone photo through the same tool and it barely moves — 25%. Same software, wildly different results, and neither number means the tool did a good or bad job. It means your two files started in very different shape. We pulled the real medians from more than 150,000 compression jobs on xconvert over a 30-day window (mid-June to mid-July 2026) and lined every format up, biggest shrinker to smallest. The production numbers below are ours; the format science that explains them is cited to MDN, Google, and the WAV/HEIF specs.
Quick answer: Across 150,000+ real jobs, the typical (median) size cut ranges from ~85% for TIFF scans down to ~25% for iPhone HEIC photos, with most everyday formats landing between 45% and 70%. The pattern is simple: uncompressed or wasteful formats (TIFF, BMP, PNG, image-heavy PDFs) shrink the most; already-efficient modern formats (HEIC, WebP, MP4) shrink the least, because their compression is already done. A big percentage isn’t proof of a better tool — it’s proof your original had a lot of slack. SVG barely compresses at all, because it’s vector text, not a bitmap.
Jump to a section
- The real numbers: median shrink by format
- Why some formats shrink 85% and others barely 25%
- The counterintuitive truth: a big shrink means a wasteful original
- What to expect for your file
- How to compress your files on xconvert
- FAQ
The real numbers: median shrink by format
This is the table competitors’ blogs can’t publish, because they don’t run the converter. Each row is the median size cut across successful jobs for that format, ranked from the biggest shrinker to the smallest.
| Format | Median size cut | Jobs measured | Why it shrinks this much |
|---|---|---|---|
| EPS | ~88% | 754 | PostScript + embedded bitmap preview |
| TIFF | ~85% | 10,187 | Usually stored uncompressed |
| JPEG | ~78% | 6,807 | Saved at near-max quality (big headroom) |
| ~72% | 1,382 | Full-resolution embedded images / scans | |
| PNG | ~67% | 5,982 | Lossless — never discards detail |
| GIF | ~58% | 26,210 | 256-color LZW; inefficient for photos |
| BMP | ~57% | 368 | Uncompressed raster |
| WebP | ~52% | 6,066 | Efficient, but often saved high-quality |
| MP3 | ~50% | 13,867 | Already lossy; often re-encoded from 320 kbps |
| WAV | ~50% | 6,515 | Uncompressed PCM |
| MP4 video | ~46% | 18,838 | Already compressed with an efficient codec |
| Audio (all formats) | ~43% | 23,780 | Pool includes efficient AAC / Opus |
| HEIC | ~25% | 8,093 | HEVC (H.265) — already highly efficient |
| SVG | ~0% | 177 | Vector XML text, not a bitmap |
Across all video formats (via the general video compressor), the median cut is ~49% over 13,230 jobs.
How we measured. These are medians — the typical result, not the best or worst — from successful jobs on xconvert’s format-specific and general compressors over a 30-day sample (mid-June to mid-July 2026), more than 150,000 jobs in total. “Size cut” is 1 − (output size ÷ input size). The table breaks out the highest-volume formats; a handful of lower-volume formats roll into the general tools and aren’t listed separately. Your result will vary — sometimes a lot — with the source file, its original quality, and the settings you choose. Treat these as expectations, not guarantees.
Why some formats shrink 85% and others barely 25%
The spread isn’t random. It tracks one thing: how much redundancy was still sitting in the file when you uploaded it.
Uncompressed and lossless formats have the most to give (TIFF, BMP, PNG, WAV). MDN’s image guide is blunt about it — “Most TIFF files are uncompressed,” and a BMP is an “uncompressed raster image, resulting in large file sizes compared to png or jpg.” A 24-megapixel scan saved as uncompressed TIFF is enormous and almost entirely redundant, so real compression carves off ~85% (our biggest high-volume result, across 10,000+ jobs). PNG is different: it does compress, but losslessly — it never discards a single pixel of detail even when your eye couldn’t tell — so converting a screenshot or logo to a lossy format, or simply re-optimizing it, typically drops ~67%. Audio has the same story: a WAV stores uncompressed linear PCM that “retains all of the samples,” which is exactly why “uncompressed WAV files are large.” (EPS edges TIFF at ~88%, but on a far smaller sample of 754 jobs, and for its own reason: an EPS often wraps a high-resolution bitmap preview around its PostScript, and that preview is where the bloat lives.)
Already-efficient modern formats have the least (HEIC, WebP, MP4). At the other end, HEIC barely budges — a ~25% median, the smallest in our data, and that’s exactly what should happen. HEIC stores images with HEVC (H.265) compression, the same modern codec used for video, and is already up to 50% smaller than JPEG at the same quality. Your iPhone did the hard compression the moment it saved the photo; there’s little redundancy left to remove. WebP is similar — Google’s own study puts lossy WebP 25–34% smaller than JPEG, and lossless WebP ~26% smaller than PNG — so it starts efficient and only gives back ~52%. MP4 video (~46%) follows the same codec logic: it’s already compressed with an efficient video codec, so re-encoding reclaims roughly half, not 85%.
Lossy formats saved with headroom shrink more than you’d expect (JPEG, MP3, PDF). Here’s the surprise in our data: JPEG — a lossy format that’s supposedly “already compressed” — still shrinks ~78%. The reason isn’t the format; it’s how people save it. Phones and editors routinely export JPEGs at near-maximum quality, which leaves enormous headroom, so re-encoding at a sensible quality removes most of it with little visible change. MP3 tells the same story (~50%): a track ripped at 320 kbps has plenty to give when re-encoded to a still-good bitrate. And PDFs (~72%) are often stuffed with full-resolution images — especially scanned PDFs, where every page is a photo — so downsampling those images does the heavy lifting. If a scanned PDF is bafflingly large, why your scanned PDF is so big breaks down the mechanics. And if you’re weighing which image format to keep, our PNG vs WebP vs JPG comparison covers the quality trade-offs behind these numbers.
Vector files don’t compress like images at all (SVG). SVG is the outlier that proves the rule: ~0% shrink. An SVG isn’t a grid of pixels — it’s an XML-based vector format, “text files containing source code” that describe shapes and lines. A raster image compressor has nothing to squeeze. To make an SVG smaller you minify the code or gzip it in transit, not run it through an image compressor.
The counterintuitive truth: a big shrink means a wasteful original
It’s tempting to read the table as a leaderboard — TIFF 85%, HEIC 25%, so the tool must “work better” on TIFF. It doesn’t. The percentage is a verdict on your original file, not the compressor. A huge cut means the source was carrying a lot of waste: uncompressed pixels, over-provisioned quality, redundant scan data. A small cut means the format already did its job before the file ever reached us.
So two honest takeaways. First, don’t judge a compressor by its headline percentage — judge it by whether it hits the size you need at a quality you accept. Second, if your HEIC or WebP file barely shrinks, you weren’t cheated — that’s the format working as designed. The real wins are hiding in scans, screenshots, and raw exports, not in files a modern codec already squeezed.
What to expect for your file
- Scans, faxes, and photo-heavy PDFs — your biggest wins, commonly 70–85% off. A giant scanned PDF isn’t a bug; it’s raw page images.
- Screenshots, logos, and graphics (PNG) — big cuts (~67%), especially if you don’t truly need lossless.
- Camera photos saved as JPEG — ~78% is normal when the original was high-quality, and the quality stays close.
- iPhone photos (HEIC) — expect only ~25%. To go smaller, reduce the pixel dimensions rather than pushing quality lower.
- Animated GIFs — around ~58%, but converting to MP4 shrinks them far more; GIF’s 256-color format is inefficient for anything photographic or moving.
- Music (MP3) and voice memos — ~50% if they started at a high bitrate. A raw WAV has even more room, but converting it to MP3 or AAC saves far more than keeping it as WAV.
- Video (MP4) — ~46–49%; the codec and target bitrate matter more than the tool.
- Web images (WebP) — modest (~52%); they’re efficient to begin with.
- SVG — don’t bother running it through an image compressor; minify the code instead.
How to compress your files on xconvert
Whatever you’re shrinking, the flow is the same: upload, choose how you want to control the size, download.
- Open the right compressor. Use the image compressor for JPG, PNG, HEIC, WebP, TIFF, BMP, GIF, and EPS; the video compressor for MP4 and other video; or the audio compressor for MP3, WAV, and other audio.
- Add your file. Click Upload (on the video compressor the button reads + Add Files) to bring it in from your computer, Google Drive, or Dropbox.
- Open Advanced Options (the gear icon) to control how much it shrinks.
- Pick how to hit your target. Leave it on Target file size (%) (marked “Best”) to let the tool balance size and quality, choose Specific file size to hit an exact number of MB, or set Image Quality (%) (images), a Custom Bitrate or Variable Bitrate (audio), or Constant Quality (video) yourself.
- Compress and download. Click Compress and save the result.
Every file you upload is processed on our servers and deleted automatically a few hours later — nothing is kept.
FAQ
How much smaller can I actually make my file?
It depends heavily on the format and how the original was saved. Across 150,000+ jobs the median cut ran from ~85% (TIFF) to ~25% (HEIC), with most everyday files landing between 45% and 70%. These are medians — your file could do better or worse. To guarantee a size, use Specific file size and enter the exact MB you need.
Why did my iPhone (HEIC) photo barely shrink?
Because HEIC is already one of the most efficient image formats there is — it uses HEVC (H.265) compression and is up to ~50% smaller than JPEG straight out of the camera. Our median HEIC cut is only ~25%, and that’s expected: your phone already removed most of the redundancy. To shrink it further, reduce the pixel dimensions rather than pushing quality lower.
Why does a JPEG shrink so much if it’s already compressed?
Because most JPEGs are saved at near-maximum quality, which leaves a lot of headroom. Re-encoding at a sensible quality reclaimed a ~78% median in our data with little visible change. The big percentage reflects a wastefully high-quality original, not a magic tool.
Which formats compress the most?
Uncompressed and lossless ones. TIFF scans top the reliable list at a ~85% median, followed by JPEGs saved at high quality (~78%), image-heavy PDFs (~72%), and PNG graphics (~67%). These formats either store raw pixels or never discard detail, so there’s the most to remove.
Does a bigger compression percentage mean a better tool?
No. A larger cut means your original was less efficient — uncompressed pixels, over-high quality, or redundant scan data. A small cut (like HEIC’s ~25%) means the format already did the work. Judge a compressor by whether it hits your target size at an acceptable quality, not by the headline percentage.
Why won’t my SVG compress?
An SVG is vector code, not a bitmap — an XML text file describing shapes — so a raster image compressor has nothing to squeeze (our median was ~0%). To make an SVG smaller, minify the code or serve it gzipped, rather than running it through an image compressor.
Are these numbers guaranteed for my file?
No — they’re medians over a 30-day, 150,000+ job sample, and individual results vary widely with the source file and the settings. Use them to set expectations. When you need a specific size, the Specific file size option lets you set the exact target directly.
Sources
Last verified 2026-07-16.
- Production figures: xconvert production data, 30-day sample (mid-June to mid-July 2026) — medians of output size ÷ input size across 150,000+ successful compression jobs. Individual results vary with the source file.
- MDN — Image file type and format guide — “Most TIFF files are uncompressed”; BMP is an “uncompressed raster image, resulting in large file sizes”; PNG uses lossless compression; GIF uses LZW on an 8-bit (max 256-color) palette; lossless WebP is “typically 26% smaller than the same images in PNG”; JPEG is lossy (DCT); SVG is an XML vector format of “text files containing source code.”
- Google for Developers — WebP Compression Study — lossy WebP is 25–34% smaller than JPEG at the same SSIM (file-size ratios 0.66–0.75).
- Wikipedia — WAV — the common WAV form is uncompressed LPCM that “retains all of the samples”; “uncompressed WAV files are large” (cites the Microsoft/IBM RIFF spec).
- Cloudinary — JPEG vs HEIC — HEIC is “part of the High Efficiency Video Compression (HEVC) standard” and “typically up to 50% smaller than JPEGs at the same quality level.”