How do I compress a PDF to exactly 100KB?

Set 100 as the target and let the search run - though "exactly" is not what you want. Upload caps are maximums, so the goal is the largest file that still fits under 102,400 bytes, which is what the search finds; landing at 99 KB satisfies the rule with a margin, and is a better outcome than 100.0 KB, not a worse one. On a multi-page scan a 100 KB cap will almost certainly require lowering the page resolution as well as the quality, so leave that option ticked or the target may be unreachable. If the portal counts a kilobyte as 1,000 bytes, aim at 97 KB to stay clear of the discrepancy.

Will compressing to a size limit make my PDF unreadable?

It depends entirely on how much budget each page gets. A one-page certificate at 100 KB has plenty to work with and will look fine. Ten pages sharing the same 100 KB gives each page about 10 KB, which forces a very low resolution and will be hard to read. The result panel always reports the resolution it settled on in DPI, so you can judge before sending the file: roughly 150 DPI is comfortable for reading, 100 DPI is usable, and below about 70 DPI small print starts to disappear. If the number looks too low, the fix is fewer pages rather than more compression.

Why can a PDF not be compressed below a certain size?

Two independent floors. First, the container itself costs bytes - about 330 per page plus a fixed overhead - so a fifty-page document has a hard floor of roughly 17 KB before a single pixel is drawn. Second, below a certain number of bits per pixel JPEG stops producing anything meaningful, so there is a practical limit to how little a readable page can cost. When a target sits under those floors the tool reports that it could not be met instead of pretending otherwise, and the way through is to reduce the page count, not the quality.

How to Compress a PDF to 100KB (or Any Size Limit)

The upload box says "PDF, maximum 100 KB". Your scanned certificate is 2.8 MB - twenty-eight times too big. The portal will not negotiate, it will not tell you how to fix it, and the browse button simply refuses the file. Unlike a photo, a PDF gives you no obvious dial to turn.

What a 100 KB limit means in bytes

A kilobyte is almost always 1,024 bytes rather than 1,000, so a 100 KB cap means 102,400 bytes. That is what Windows, macOS and most upload validators use, and it is the number worth working to.

A minority of portals do the arithmetic in round decimal thousands, making the same cap 100,000 bytes - about 2.4 per cent tighter. That gap is exactly the margin a tool aiming to land just under 102,400 will be sitting in. If a file that looks comfortably inside the limit is refused anyway, this is the usual reason: set the target a couple of kilobytes lower and compress again.

Where a PDF actually spends its bytes

A PDF is a container. Some of its bytes describe the document itself - the page tree, the cross-reference table, the object that says how big each page is - and the rest hold the content. The container part is small but not free, and it is why very small targets fail.

Measured on real output, a rebuilt PDF costs roughly 330 bytes per page plus about 600 bytes fixed: a three-page file spends about 1,600 bytes on structure, a ten-page file about 3,900. So a ten-page document cannot be squeezed into 2 KB no matter how bad the pages look, because the empty container already costs twice that. Everything left over after the structure is the real budget for the pages.

That leftover is spent one way: the pages are re-drawn as compressed JPEG images. Two things then decide the size - the JPEG quality, and how many pixels each page is drawn with. The second one multiplies everything, because halving the width and height quarters the pixel count.

A worked example: spending a 200 KB budget

Turn the target into an allowance per pixel and the whole problem becomes visible. Take a three-page scan with a 200 KB cap.

200 KB is 204,800 bytes. Structure takes about 1,600, leaving roughly 203,200 for the pages.
Split three ways, that is about 67,700 bytes per page - or 541,600 bits.
Draw an A4 page at 115 DPI and it is about 952 x 1347 pixels, roughly 1,280,000 of them. Each pixel gets 541,600 / 1,280,000, about 0.42 bits.
Draw the same page at 75 DPI and it is about 620 x 875 pixels, roughly 540,000. Now each pixel gets about 1.0 bits - more than twice as much.

JPEG holds up well at around 1 bit per pixel and falls apart well below it, so the second version is the readable one. Nothing was added to the budget; the same 200 KB was simply shared between fewer pixels. That is the entire trick, and it is why a tool that only lowers the quality slider will hand you a blurry mess while insisting it did what you asked.

It also predicts the answer. Given that scan and a 200 KB target, the Compress PDF to Size tool settles on about 75 DPI - the resolution the arithmetic above points at - and produces a 198.7 KB file.

Hitting the number

Guessing at a quality setting and re-checking the file size is slow, and it is how people end up with a 71 KB file when they were allowed 100. The tool searches instead: it renders your pages once, then hunts for the highest quality whose rebuilt PDF still fits, and lowers the page resolution only when quality alone cannot get there.

Drop your PDF in. If it is password-protected you will be asked for the password first, and it is used in your browser only.
Type the cap in the target box, or use a preset - 100 KB, 200 KB, 300 KB, 500 KB, 1 MB and 2 MB cover most forms.
Leave the resolution option ticked. It only acts when the target cannot be reached at the current resolution, and it is what makes a 100 KB cap possible on a multi-page scan.
Press compress and read the result panel: the exact byte count against your budget, the page count, the resolution in DPI it settled on, and the quality it used.

Results land just under the cap rather than far below it - typically 98 to 99 per cent of the budget - because every unused byte is detail you were entitled to and did not get. If your PDF already fits, it is handed back untouched rather than rebuilt, so nothing is lost and its text stays selectable.

What you give up

Reaching a fixed byte target means the pages become images, so text stops being selectable or searchable. Every size-target compressor makes this trade, and it is worth knowing before you send the file somewhere that might need to copy text out of it.

It also means the approach suits some documents far better than others. Scans, photographed certificates and image-heavy reports shrink enormously and lose little that matters. A plain text PDF is a poor candidate: it is already small and efficient, and rasterising it can make it larger while destroying its text. If you have no exact number to hit and just want a smaller file, the plain Compress PDF tool is the better fit - it optimises for quality per byte instead of chasing a cap.

When the target is out of reach

Sometimes the number is simply not achievable at a readable quality. Three things help, in order of how much they buy you:

Send fewer pages. The budget is shared between them, so dropping a page makes every remaining page sharper. Use Remove Pages to cut what the form does not need, or Split PDF if the portal accepts separate uploads.
Check whether an image is really wanted. Many forms asking for a 20 KB signature or a 50 KB photograph want a JPG, not a PDF - in which case Compress Image to Size does the job with the container overhead removed entirely.
Raise the target slightly if the form allows a range. The difference between 100 KB and 150 KB is often the difference between legible and not.

When a target genuinely cannot be met, the tool says so plainly and still offers the smallest file it managed, rather than silently handing back something over the limit.

Why this runs in your browser

Documents that come with a byte limit attached are rarely trivial ones - they are certificates, identity documents, signed forms and bank statements. Every step here happens on your own device: the file is read locally, its pages are rendered and re-encoded by your browser, and the download comes from memory. Nothing is uploaded, and no copy is left on anyone else's disk.

Open Compress PDF to Size, name your limit, and let it find the resolution and quality that fit.

Frequently asked questions

How do I compress a PDF to exactly 100KB?: Set 100 as the target and let the search run - though "exactly" is not what you want. Upload caps are maximums, so the goal is the largest file that still fits under 102,400 bytes, which is what the search finds; landing at 99 KB satisfies the rule with a margin, and is a better outcome than 100.0 KB, not a worse one. On a multi-page scan a 100 KB cap will almost certainly require lowering the page resolution as well as the quality, so leave that option ticked or the target may be unreachable. If the portal counts a kilobyte as 1,000 bytes, aim at 97 KB to stay clear of the discrepancy.
Will compressing to a size limit make my PDF unreadable?: It depends entirely on how much budget each page gets. A one-page certificate at 100 KB has plenty to work with and will look fine. Ten pages sharing the same 100 KB gives each page about 10 KB, which forces a very low resolution and will be hard to read. The result panel always reports the resolution it settled on in DPI, so you can judge before sending the file: roughly 150 DPI is comfortable for reading, 100 DPI is usable, and below about 70 DPI small print starts to disappear. If the number looks too low, the fix is fewer pages rather than more compression.
Why can a PDF not be compressed below a certain size?: Two independent floors. First, the container itself costs bytes - about 330 per page plus a fixed overhead - so a fifty-page document has a hard floor of roughly 17 KB before a single pixel is drawn. Second, below a certain number of bits per pixel JPEG stops producing anything meaningful, so there is a practical limit to how little a readable page can cost. When a target sits under those floors the tool reports that it could not be met instead of pretending otherwise, and the way through is to reduce the page count, not the quality.