Two Numbers That Tell Very Different Stories
When people talk about internet performance, they usually focus on one number: how fast is my connection? But "fast" is ambiguous. A connection can move a lot of data per second (that's throughput) while still feeling sluggish on interactive tasks (that's latency). Understanding the difference between latency and throughput is the key to diagnosing why your internet sometimes feels broken even when your speed test looks great.
What Is Latency?
Latency is the time it takes for a single packet of data to travel from your device to a server and back. It's measured in milliseconds (ms), and lower is always better. You'll sometimes hear it called "ping time" because the ping command is the most common way to measure it.
For context:
- < 5 ms: Excellent — typical of a fiber connection to a nearby server
- 5–20 ms: Great for almost everything, including competitive gaming
- 20–50 ms: Fine for most uses; video calls might show minor delays
- 50–100 ms: Noticeable lag in real-time applications
- > 100 ms: Problematic for gaming, VoIP, and interactive sessions
Latency is determined by physics (the speed of light in fiber is finite), network routing, and the number of hops between you and the destination. The IETF's network performance standards define the metrics used to measure these constraints. You can't throw bandwidth at a latency problem — it's a fundamentally different constraint.
What Is Throughput?
Throughput is the actual rate at which data successfully transfers across a connection. It's what your speed test measures. If your plan advertises 500 Mbps and your test shows 480 Mbps, your throughput is 480 Mbps.
Throughput depends on bandwidth (your connection's capacity), but also on network congestion, protocol overhead, hardware limitations, and the performance of the server on the other end. It's a measure of useful data delivered, not just raw capacity.
The Key Differences Between Latency and Throughput
| Characteristic | Latency | Throughput |
|---|---|---|
| What it measures | Time for one round trip | Volume of data per second |
| Unit | Milliseconds (ms) | Megabits per second (Mbps) |
| Analogy | How fast one car travels | How many cars fit on the road |
| Affected by | Distance, routing, hops | Bandwidth, congestion, hardware |
| Critical for | Gaming, VoIP, video calls | Streaming, downloads, backups |
| Fixable with more bandwidth? | No | Sometimes |
Why Both Matter
Here's the thing most people miss: a 1 Gbps connection with 200 ms latency will feel worse for video calls than a 50 Mbps connection with 5 ms latency. Throughput is meaningless if every interaction has a quarter-second delay baked in.
Conversely, ultra-low latency won't help you download a 50 GB game update in any reasonable time. You need both numbers to be healthy, and which one matters more depends entirely on what you're doing.
Real-World Examples
Online Gaming
Latency is king. A competitive FPS player on a 25 Mbps connection with 8 ms ping will demolish a player on a 1 Gbps connection with 80 ms ping. Games send tiny packets of data constantly — your position, your aim, enemy movements. Each millisecond of delay puts you behind. Throughput barely matters because the data volume is minuscule.
4K Streaming
Throughput is king. Netflix recommends 15 Mbps for 4K (services like Disney+ ask for 25 Mbps). Latency is almost irrelevant because the video buffers ahead — you're not interacting in real time. A 100 ms ping won't degrade your viewing experience at all.
Video Conferencing (Zoom, Teams)
Both matter. You need enough throughput to handle incoming and outgoing video (typically 10–20 Mbps), but latency above 150 ms creates awkward conversational overlaps and audio desync. Zoom's own system requirements detail the exact bandwidth and latency thresholds their platform needs. Jitter — variation in latency — is particularly annoying here because it causes audio to stutter.
Large File Transfers
Throughput dominates. Whether you're backing up to the cloud or pulling a 100 GB dataset from a server, total transfer time depends on throughput. Latency adds a small overhead per packet, but with modern TCP window sizes and protocols like QUIC, its impact is minimal for bulk transfers.
How to Measure Both
A standard speed test measures throughput. For latency, use the ping command in your terminal or check the latency line on a detailed speed test result. For the most accurate picture, test to a server close to you (to isolate local network performance) and to a distant one (to see the effect of routing and distance).
Which Should You Optimize For?
If your connection feels slow but your speed tests look fine, latency is likely your bottleneck. Check your router, look for bufferbloat issues, or consider upgrading your networking equipment. NIST's networking research division publishes standards on measuring and optimizing network performance. If downloads crawl and streams buffer, throughput is the issue — you may need a faster plan or need to diagnose what's consuming your bandwidth.
Understanding the difference between latency and throughput stops you from throwing money at the wrong problem. A 2 Gbps plan won't fix your Zoom calls if your latency is 300 ms — but a better router might.