What really limits your file transfer speed

A file transfer runs at the speed of its slowest link. In practice that is usually the sender's upload speed, which on many home connections is a fraction of the download speed printed on the bill. After that, the usual suspects are the receiver's download speed, Wi-Fi signal (band, distance, walls), any server or relay the data passes through, and simple confusion between megabits and megabytes, which makes every transfer look eight times slower than expected. Measure your upload speed, do the arithmetic, and you will know within a minute whether a transfer is slow or just big.

The one formula you need

Transfer time ≈ file size in megabits ÷ the slowest link in Mbps, plus 5 to 15% for protocol overhead.

To get megabits, multiply megabytes by 8. To find the slowest link, take the lowest of: the sender's upload speed, the receiver's download speed, and anything in between (a cloud service's throughput, a relay server, a weak Wi-Fi hop).

Worked example

You send a 2 GB video. Your home plan is "300 Mbps" download but only 20 Mbps upload. The recipient has 500 Mbps download.

1. 2 GB = 2,000 MB = 16,000 megabits.
2. Slowest link = your 20 Mbps upload.
3. 16,000 ÷ 20 = 800 seconds, about 13 minutes, so expect 14 to 15 minutes with overhead.

The recipient's 500 Mbps does not help at all. If your upload were 200 Mbps, the same file would take under 2 minutes. Our transfer time calculator does this arithmetic for any size and speed.

Mbps vs MB/s: the factor of eight

Internet speeds are sold in megabits per second (Mbps, lower-case b). File sizes, and the progress bars in most browsers and operating systems, use megabytes (MB, upper-case B). One byte is eight bits.

Figures use decimal units (1 GB = 1,000 MB) and no overhead; real transfers run somewhat slower. If a progress bar shows 11 MB/s on a "100 Mbps" connection, nothing is wrong. That is close to the maximum.

Upload vs download: why your plan's big number misleads

Cable and DSL connections are typically asymmetric. Providers allocate most of the capacity downstream, because households download far more than they upload. Even regulators reflect this: in 2024 the US Federal Communications Commission raised its benchmark for broadband to 100 Mbps download and 20 Mbps upload, a 5-to-1 ratio. Fibre plans are more often symmetric, and mobile upload speeds vary widely with signal.

For a file transfer, the sender's upload is what counts. So the most useful thing you can learn about your connection is its upload figure. If it is 10 or 20 Mbps, sending a large video will take a while no matter which service you use.

Services in the middle: upload twice, queues and relays

With a cloud-based service, the file crosses the internet twice: the sender uploads it completely, then the recipient downloads it. The recipient can usually only start once the upload has finished, so total time is roughly upload time plus download time, bounded by the service's own throughput.

Some free tiers also deprioritise free users. At the time of writing (October 2026), Smash's pricing page says that on the free plan, transfers over 2 GB "join a queue behind Pro transfers", and its FAQ says paid plans get "priority delivery". That is a reasonable way to fund a free service, but it means the time you wait is not purely a function of your connection. Check the plan details of whichever service you use before blaming your internet.

Direct peer-to-peer transfers avoid the double trip: the data goes from the sender's device to the receiver's as it is read, so time is roughly the slower of the two connections. On TransferFiles.net, for example, we impose no speed cap and the file is not stored on a server. The caveat is that when a strict firewall blocks a direct connection, traffic is relayed through a TURN server (still encrypted end to end), and then the relay's capacity is one more link in the chain. Both people also have to stay online for the whole transfer. If you need the recipient to download later, a cloud service is the right tool despite the extra step.

Wi-Fi: band, distance and congestion

Your internet plan is irrelevant if the Wi-Fi hop between your device and your router is slower than it. Common causes:

• Band. 2.4 GHz reaches further and gets through walls better, but it is slower and crowded with neighbours' networks, Bluetooth and microwave ovens. 5 GHz is much faster at short range. 6 GHz (Wi-Fi 6E and Wi-Fi 7) is faster still and less congested, but needs compatible devices and has the shortest range.
• Distance and walls. Speed falls off quickly with range. A laptop two rooms away on 5 GHz can be slower than the same laptop on 2.4 GHz.
• Sharing. Wi-Fi capacity is shared. A 4K stream, a game download or a cloud photo backup on another device takes bandwidth from your transfer.
• Old devices. An older phone or laptop may only support older Wi-Fi standards, capping its speed regardless of the router.

For large transfers from a computer, an Ethernet cable is the simplest upgrade: consistent speed and no interference. For phones, move close to the router and pause background backups.

Local transfers skip your internet connection

When both devices are in the same building, you can often avoid your internet connection entirely. Tools such as LocalSend, AirDrop, Quick Share and SMB network shares send data across your router, so your 20 Mbps upload no longer matters; the limit becomes your Wi-Fi or Ethernet. That is why copying a big folder between two laptops in the same room can be ten times faster through a local tool than through a cloud link. Our guide to peer-to-peer file sharing compares the local options.

How to find your bottleneck: a five-minute test

1. Test the sender's upload. Run a speed test (Speedtest by Ookla, Cloudflare's speed.cloudflare.com, or fast.com with "more info" expanded) on the device that will send, on the same network it will use. Note the upload figure.
2. Test the receiver's download the same way.
3. Test near the router, then where you actually sit. A big difference means Wi-Fi signal, not your plan, is the issue.
4. Do the arithmetic with the lower of the two numbers (or use the calculator). If the real transfer is within about 20% of that, everything is working as expected.
5. If it is far slower, suspect something in the middle: the service's speed or queue, a relay, a VPN, or a congested network. Try another network (a phone hotspot is a quick test), turn off the VPN, or try a direct transfer.

Quick fixes, in order of payoff

• Move the sending device to Ethernet or right next to the router, on 5 GHz or 6 GHz.
• Pause cloud backups, streaming and game downloads on the network.
• Let whoever has the faster upload do the sending.
• If both devices are nearby, use a local transfer and skip the internet.
• Turn off VPNs for the duration if your organisation allows it; they add a detour and their own speed limit.
• Compress only what compresses. Zipping documents can shrink them a lot; zipping video or JPEGs gains almost nothing (see how to zip files).
• On restrictive networks where transfers will not start at all, see file transfer blocked on a network.

Sources

• Smash pricing and FAQ
• FCC 2024 Section 706 Report (FCC 24-27)