You drop a 10GB master into the browser. The bar climbs. 70%, 85%, 92%... then a flash of white. "Upload failed." You refresh, click upload again, and the bar starts at zero.

That's the failure that eats evenings. A long upload dies near the end, and because the tool can't remember what already made it across, you start from the top. Ten gigabytes of uplink time, gone. And the closer you got to 100%, the worse it feels.

What "resumable" actually means

A resumable upload is a transfer that can pick up where it left off.

The trick is chunking. Instead of sending your file as one long stream, the client slices it into many small pieces — think 20MB parts, not one 10GB blob. Each part gets its own progress. When a part lands, it's recorded on the server.

If the connection drops at 92%, the parts that already arrived stay arrived. When you come back, the client asks the server "how far did you get?", finds the last good part, and sends only what's missing. No restart. The part that made it doesn't get re-sent — just the one that fell out of the post.

Think of it like posting a long book one chapter at a time. If one envelope goes missing, you resend that chapter, not the whole book.

There's an open standard behind this called tus (The Upload Server), a lightweight protocol that says "here's my file, tell me the byte offset you have, and I'll continue from there." Closing the tab or losing the connection lets you continue instead of starting over. If the service speaks tus or does its own chunking, your upload is resumable. If it just streams one long POST and the tab or network blinks, you're back to square one.

Why it's table stakes at 100GB+

A 2GB send that fails is an annoyance. A 100GB delivery that fails at 88% is a deadline problem — that's 88GB, roughly ten hours on a 20Mbps uplink, to send all over again. The bigger the file and the longer the transfer, the more likely something blinks mid-flight: a laptop sleeps, a router reboots, the ISP hiccups. Without resumption, every single blink means a full restart.

The real cost isn't the bytes, it's the time you pay to re-send them. And the time grows with the file, so the penalty is worst exactly where your files are biggest.

For client delivery at 100GB+, resumable upload isn't a nice-to-have. It's the difference between "one upload, one coffee break" and "three all-nighters and a re-upload."

Which services support it

Only some services say so in writing:

The big names — WeTransfer, Smash, and Dropbox Transfer — don't document resuming a browser upload from a drop. WeTransfer's own troubleshooting page says sleep mode cancels an upload and says nothing about resuming; Smash's help says a transfer stuck at 100% has to be started again; Dropbox doesn't state it for Transfer. (Dropbox's API does support resumable upload sessions, and its desktop app recovers large files — but that's a different path from the browser you'd use for a client delivery.)

The practical test is simple: drop a large file, kill your network at 50%, wait, and reconnect. If the bar jumps back to ~50%, you've got resumption. If it resets to zero, you don't. Ten seconds of testing beats a marketing page.

Make resumption a requirement

You can't change your ISP, and you can't stop your laptop from sleeping. But you can change the tool. A service that chunks your file, verifies each part, and keeps progress across a dropped connection turns a 92% death into a five-second pause.

If you're sending 100GB+ to clients, make resumption a requirement before you pick a transfer tool — not an afterthought when the first upload dies. Check it once, and stop losing evenings.

See how a delivery-first service handles big sends: Beautifully Delivered. And for the full failure map, Why Do Uploads Fail? covers the other reasons your upload dies.

Start free →