A reliable baseline for a single HD video call is about 5 Mbps down and 3 Mbps up, according to NetForecast's work-from-home video conferencing study. That number only tells half the story: latency, jitter and packet loss, as detailed in ITU-T P.940, often decide whether a call feels smooth or choppy. Start by testing your upload speed and latency, and plug in an Ethernet cable if you can.
TL;DR:
- Small group calls need 8 to 10 Mbps down and 3 to 4 Mbps up; gallery meetings need 10 to 25 Mbps down.
- Aim for round trip time below 200 milliseconds and packet loss at or below 3%; higher loss can cause freezes and audio dropouts.
- Reserve 1.5 to 2 times your expected upload capacity for one camera; several simultaneous household calls increase both upload and download demand.
- Screen sharing adds 1 to 2 Mbps in both directions, and 1080p calls need extra headroom; turn off your camera if presenting strains upload.
- Repeated low upload speeds suggest a plan limit, while latency or loss spikes point to congestion or equipment trouble; compare wired and wireless tests.
Table of Contents
- Recommended speeds for every meeting type
- Why upload speed decides whether others see you clearly
- The quality factors that matter more than raw speed
- How to test your connection before the next call
- Quick fixes to try before changing anything else
- When to upgrade your plan or your hardware
- Why published bandwidth numbers never quite match
- What years of bad calls taught me about bandwidth
- Check your real numbers with GrandSpeed
- FAQ
- Sources
Recommended speeds for every meeting type
Your bandwidth needs change with the size of the call, how many video feeds you are sending or watching, and whether anyone is sharing a screen. A one-on-one call is the lightest load: 5 Mbps down and 3 Mbps up covers HD video with headroom, per NetForecast's baseline. Small group calls need more on the download side because you are receiving several video streams at once, even though your own upload stays close to the one-on-one number unless you are the one presenting.
Large meetings with dozens of participants rarely increase your upload needs much, since most platforms cap what any single sender pushes out. Download demand rises instead, especially in gallery view where many small thumbnails update at once.
Typical targets by scenario:
- One-on-one HD call: 5 Mbps down, 3 Mbps up
- Small group (3 to 8 people): 8 to 10 Mbps down, 3 to 4 Mbps up
- Large meeting or webinar (gallery view): 10 to 25 Mbps down, 3 Mbps up
- Screen sharing added to a call: add 1 to 2 Mbps in both directions
- 1080p send or receive: expect higher sustained bitrate than 720p, so build in extra headroom on both ends
Screen sharing adds load on top of your existing video feed rather than replacing it, which is why calls often stutter the moment someone starts presenting a slide deck.
Why upload speed decides whether others see you clearly
Most home internet plans are asymmetric: download speed is generous, upload speed is not. That asymmetry matters here because your outgoing video depends entirely on your upload connection. A slow upload link does not just affect you, it affects everyone watching your feed.
Large meetings change the math slightly. Adding more participants mostly increases your receive bandwidth, since you are pulling in more thumbnails, while your send bandwidth stays tied to your own camera and audio.
When upload is tight, conferencing apps use adaptive codecs that automatically drop your resolution or frame rate to keep the connection alive instead of freezing entirely.
Practical guidance:
- Keep 1.5 to 2 times your expected upload need as headroom for a single camera
- Treat upload as the limiting factor in multi-person households
- Watch for blurry or pixelated outgoing video as the first sign of upload strain
The quality factors that matter more than raw speed
Latency, jitter and packet loss often matter more than your plan's advertised speed. Round-trip time under 200 milliseconds keeps conversation feeling natural, per Microsoft's Teams quality monitoring guidance; anything higher introduces the awkward talk-over-each-other delay common on long-distance or satellite connections.
A stable connection with video packet loss at or below 3% typically holds up under ITU-T P.940's quality model, while higher loss rates cause visible freezing and audio dropouts regardless of your download speed.
Jitter, the variation in how quickly packets arrive, forces apps to buffer incoming data, and an overloaded router buffer (bufferbloat) adds delay on top of that. Wired Ethernet avoids most of the jitter that Wi-Fi interference introduces.
What to prioritize when diagnosing a bad call:
- Round-trip time under 200 ms for natural conversation
- Packet loss under 3% to avoid freezing and audio gaps
- Low jitter, which usually means switching from Wi-Fi to Ethernet
- No bufferbloat, which means your router is not queuing packets under load
A stable 5 Mbps down and 3 Mbps up connection with low loss will outperform an unstable 100/20 Mbps line every time.
How to test your connection before the next call
Testing tells you whether you have a speed problem or a quality problem, and the two require different fixes. Run a full diagnostic over both Ethernet and Wi-Fi so you can compare the two, capturing download, upload, latency, jitter and packet loss on each.
During an actual call, check your app's own connection stats: round-trip time, sent bitrate and packet loss. If your speed test shows plenty of bandwidth but the call still stutters, the problem is almost always jitter, loss or bufferbloat rather than raw capacity.
A simple test sequence:
- Run a speed test over Ethernet, note download, upload, latency and jitter
- Repeat the same test over Wi-Fi from the room where you take calls
- Join a test call and check the app's bitrate and packet loss stats
- Compare both sets of numbers to spot where the gap appears
Pro Tip: Run your test at the same time of day your calls usually happen, since shared neighborhood bandwidth or household streaming can change your numbers hour to hour.
Quick fixes to try before changing anything else
Start with the cheapest fix first: plug in an Ethernet cable. Wired connections remove most Wi-Fi jitter and interference in one step. If that is not possible, reboot your router and move your device closer to it.
Pause anything competing for bandwidth during calls. Cloud backups, large file syncs and a roommate's video stream can all eat into your upload or download without you noticing until your own call drops frames.
Fast changes that help immediately:
- Switch to wired Ethernet before anything else
- Pause cloud backups, software updates and file syncs during calls
- Skip VPNs for calls when possible, since they add latency
- Lower your in-app resolution or disable HD video if the call struggles
- Turn off your camera while sharing a screen to free up upload capacity
On the router side, enabling basic Quality of Service (QoS) settings, separating guest and main Wi-Fi networks, and keeping firmware updated all reduce the chance of one device starving your call of bandwidth.
Pro Tip: If a call improves the moment you switch to Ethernet, your issue was Wi-Fi interference, not your internet plan.
When to upgrade your plan or your hardware
If your upload speed tests low consistently across different times of day, that is a plan limitation, not a temporary glitch. Occasional packet loss or latency spikes that come and go point toward network congestion or equipment issues instead.
Households or small offices running multiple simultaneous calls need more total capacity than a single-user baseline suggests, since each active camera adds its own upload and download load.
What to check before upgrading anything:
- Consistent low upload across multiple times of day points to your plan
- A modern router, a gigabit switch, or a wired network card can remove local bottlenecks without a plan change
- Save test logs showing time of day and repeated packet loss before calling your ISP, since patterns carry more weight than a single bad test
Why published bandwidth numbers never quite match
Different sources give different numbers because they measure different things. NetForecast's study recommends 5 Mbps down and 3 Mbps up as a conservative baseline built from 95th-percentile usage across real sessions, leaving a buffer above what most calls actually consume.
ITU-T P.940 takes a different angle, modeling perceived quality from bitrate, packet loss, delay and jitter together rather than bandwidth alone. That is why a connection can meet every speed benchmark and still feel choppy.
Vendor guidance adds a third layer of variation. Microsoft's live event planning documentation lists per-stream payload ranges that shift as adaptive codecs adjust bitrate in real time, so the number an app actually uses rarely matches its published maximum.
What years of bad calls taught me about bandwidth
My rule of thumb: wired first, test from the actual spot you take calls, and keep 1.5 to 2 times your expected upload as headroom. Speed tests run from a different room than your desk tell you very little.
The most common oversight I see is an automatic backup kicking off mid-meeting, or a roommate starting a 4K stream the moment you join a call. Neither shows up until the video freezes.
— Peter D
Check your real numbers with GrandSpeed
Everything above assumes you know your actual upload, download, latency, jitter and packet loss, not just the number on your internet bill. Our Speed Test measures download, upload, latency, jitter and packet loss, so you can see where a call is likely to struggle before it happens.
Once you have real numbers, our Video bitrate planner helps you plan for a suitable resolution for calls, screen sharing or live events. Use these tools to check the numbers you gathered above rather than guessing from an advertised plan speed.
- Measure download, upload, latency, jitter and packet loss in one test
- Map your tested speed to a realistic streaming or call resolution
- Repeat tests on Ethernet and Wi-Fi to isolate the real bottleneck
| Scenario | Download target | Upload target |
|---|---|---|
| One-on-one HD call | 5 Mbps | 3 Mbps |
| Small group (3 to 8) | 8 to 10 Mbps | 3 to 4 Mbps |
| Large meeting or webinar | 10 to 25 Mbps | 3 Mbps |
For teams running production-grade events with multiple cameras, Television handles on-location livestream production and screen rental, a useful option once bandwidth planning moves beyond a single desk setup.
FAQ
What is the recommended upload speed for Teams video calls?
Microsoft's own monitoring guidance targets round-trip time under 200 milliseconds and packet loss under 2% for healthy Teams calls. A general upload baseline of 3 Mbps for a single user, drawn from NetForecast's research, is a reliable starting point for HD video conferencing.
How many Mbps do I need for video conferencing?
A single HD video call needs about 5 Mbps down and 3 Mbps up as a reliable baseline, according to NetForecast. Larger meetings or multiple simultaneous calls in one household need more on the download side to cover extra video feeds.
Is 300 Mbps fast enough for Zoom calls?
A fast download connection is generally more than enough for a video call, since upload speed and connection stability can also affect call quality. Check your upload speed and latency specifically, since a fast download tier does not guarantee a fast upload tier.
How much bandwidth does a video call use per hour?
Video call data use varies by resolution, platform and settings. Check the app's data-use information or monitor your usage to estimate what your calls consume. Our breakdown of video call data usage covers the full range by call type and quality setting.
Sources
- Internet Connection Requirements for Effective Video Conferencing to Support Work from Home
- Recommendation ITU-T P.940 (03/2025) - Computational model used for the monitoring and quality assessment of videotelephony services