Your connection speed test looks great, but the video call keeps freezing, the game rubber-bands, and pages randomly stall. The culprit is often invisible to a speed test: packet loss. A little of it degrades everything quietly, and it's one of the most common-yet-overlooked network problems.
What packet loss is
Data travels the internet in small chunks called packets. Packet loss is when some of those packets never reach their destination — they're dropped somewhere along the way. It's usually expressed as a percentage: "2% packet loss" means 2 of every 100 packets vanished.
A few lost packets sounds trivial, but the effects are outsized, because of how the internet reacts to loss.
Why a little loss hurts a lot
For reliable protocols like TCP, a lost packet must be detected and re-sent — which adds a noticeable delay and can throttle the whole connection's speed. So even small loss inflates your effective latency and tanks throughput.
For real-time protocols like UDP (calls, games, streaming), there's no time to re-send — so a lost packet is just gone, which you experience as a glitch, a freeze, or a dropped word.
| Loss level | What you notice |
|---|---|
| Under ~1% | Usually fine for most uses |
| 1–2.5% | Calls/games start to stutter |
| Over 2.5% | Noticeable problems; pages crawl |
| 5%+ | Things feel broken |
What causes it
| Cause | Where it happens |
|---|---|
| Network congestion | An overloaded link drops excess packets |
| Faulty hardware | A bad router, cable, or port |
| Wi-Fi interference | Weak signal, crowded channels |
| An overloaded server | The destination can't keep up |
| A bad hop / route | One problematic router on the path |
How to find it
This is where the network tools earn their keep:
- Ping shows loss directly — if some replies never come back, you have loss.
- MTR is the star here: it runs continuously and shows loss per hop, so you can pinpoint where packets start disappearing. Intermittent loss that a one-off test misses, MTR catches.
The key skill: find where the loss starts. Loss that appears at one hop and continues from there usually means that hop (or the link to it) is the problem. Loss only at the final hop might just be a router de-prioritising pings while forwarding real traffic fine.
How to fix it
- Your side: swap suspect cables, fix weak Wi-Fi, reduce local congestion.
- The route: if a specific hop (often a provider's) is dropping packets, it may be out of your hands — but documenting it with MTR helps when you contact your ISP/host.
- The server: if loss is at the destination, it may be overloaded — add capacity.
Why monitor for it
Packet loss is intermittent and invisible to a basic "is it up?" check — yet it degrades the experience badly. Continuous ping/network monitoring that tracks loss (and round-trip time) from multiple locations catches it when it starts, and shows whether it's global or regional.
The bottom line
| In one line | |
|---|---|
| What | Packets that never reach their destination. |
| Why it hurts | Triggers slow retransmissions (TCP) or glitches (UDP). |
| Find it | Ping for loss; MTR for where it happens. |
| Even 1–2% | Enough to wreck calls, games and streaming. |
Packet loss is the silent saboteur of "fast" connections — small percentages, big pain. Use ping to spot it and MTR to locate it, and you'll turn "the network is weird" into "hop 6 is dropping packets."
Tools: free ping and traceroute; deeper in ping, traceroute & MTR.