Bufferbloat Test

Your speed test says your broadband is fast, yet your video calls freeze and your games lag the moment anyone else in the house downloads something. That contradiction is usually bufferbloat — and this free bufferbloat test measures it directly. It checks your latency while your connection is idle, then again while your download and upload are fully busy, and turns the difference into a simple grade from A+ to F. The test runs in your browser in about 25 seconds and is part of Speed Test UK.

Most speed tests only measure an idle, unloaded line — the equivalent of timing a car on an empty road. Real life is the rush hour: a game update downloading, a cloud backup uploading, a partner on a video call, all at once. Bufferbloat is what happens to your connection in that rush hour, and it explains a large share of “my internet feels slow even though the speed is fine” complaints in UK homes.

What Is Bufferbloat?

Every router and modem has buffers — small waiting areas where data queues before being sent. Buffers are necessary: they smooth out bursts of traffic. Bufferbloat is what happens when those buffers are far too big and unmanaged. When your connection gets busy, data piles into the queue and waits there, sometimes for hundreds of milliseconds or even whole seconds. Your download speed barely changes, but everything interactive — calls, games, remote desktops, voice chat — is stuck in the queue behind a bulk download.

A household example makes it concrete. Your line tests at 70 Mbps with a tidy 12 ms ping. Someone starts a photo backup, which saturates your 18 Mbps upload. Your game traffic now queues behind megabytes of backup data inside the router’s buffer. Your ping does not rise by the 1 or 2 ms the data “should” take — it rises to 250 ms, because your packets are waiting their turn behind a queue that is seconds deep. That gap between idle latency and busy latency is exactly what this test measures.

How to Use This Bufferbloat Test

The test has three automatic phases — idle, download load, then upload load — and takes about 25 seconds in total. A few minutes of preparation makes the result far more trustworthy.

• Use a wired connection if you can. Plug a laptop into your router with an Ethernet cable. Wi-Fi adds its own variable delay, which this test cannot separate from true bufferbloat. If you must test over Wi-Fi, treat the grade as a guide and re-test wired before changing anything.
• Quiet the house for half a minute. Pause game downloads, streaming on other screens, cloud backups and video calls. Other traffic during the test will look like bufferbloat that is not your router’s fault.
• Disconnect any VPN. A VPN adds its own delay and its own queues. Test your real connection first; you can test the VPN afterwards for comparison.
• Press Start Test and leave the tab alone. Do not switch tabs — browsers throttle background tabs, which corrupts the measurements.
• Read the grade, then both increases. The table shows idle latency, latency during download, latency during upload, and your download and upload speeds. The grade is based on whichever increase is worse.

One honest note about method: browsers cannot send traditional ICMP pings, so this test times tiny HTTPS requests to Cloudflare’s edge network, which has servers in the UK, while separately saturating your line with Cloudflare’s speed-test endpoints. That is the same basic approach used by the well-known web-based bufferbloat tests, and it is accurate enough to grade a home connection reliably.

Your Results Explained: The Grades

The grade is worked out from how many extra milliseconds of latency appear when your connection is busy, compared with when it is idle. It is based on the worse of the download and upload figures, because the worse one is the one you will feel.

Two other numbers deserve a look. First, your idle latency: if it is already high (say over 50 ms) before any load, your problem may not be bufferbloat at all — distance, a mobile or satellite connection, or a line fault may be the cause, and our ping tester will help you dig in. Second, the split between download and upload: upload grades are often worse, because UK home uploads are slower, so the upload queue fills and stays full for longer. A connection can honestly earn an A for download and a D for upload; the headline grade reflects the D, because that is what your calls experience during a backup.

Why Ordinary Speed Tests Miss Bufferbloat Completely

A standard broadband speed test measures throughput and idle ping on a quiet line. It never asks the question that matters for real-time use: what happens to latency while that throughput is being used? A line can post excellent speed numbers and still be miserable for a video call, because the two problems are different — one is the width of the road, the other is the length of the queue at the junction. That is also why your provider’s support line, working from a speed result, may tell you nothing is wrong while your evening gaming session disagrees.

Bufferbloat is closely related to jitter — the variation in latency from moment to moment. As queues fill and drain, your ping swings up and down, and those swings are what make voices sound robotic. If this test gives you a poor grade, our jitter test is the natural next check to see how unstable your latency is over time.

Bufferbloat in UK Homes: Lines, Routers and Peak Time

The UK has a few patterns worth knowing. On part-fibre (FTTC) lines — fibre to the street cabinet, copper to your home — headline speeds are modest and uploads are slow, so queues build quickly and the upload side is usually where the grade is lost. Full fibre (FTTP) connections have far more headroom, which hides bufferbloat in everyday use, but a saturated gigabit line at peak time can still queue badly in a router with unmanaged buffers. The equipment matters as much as the line: the free routers supplied by UK providers — BT’s Smart Hub, Sky’s Broadband Hub, Virgin Media’s Hub range, TalkTalk and Plusnet hubs — vary in how they manage queues, and most prioritise simple throughput over latency under load. Virgin Media’s cable network has its own congestion pattern at busy times, so an evening test and a morning test on the same line can honestly produce different grades; that difference itself is useful evidence.

For impartial advice on getting the most from a UK connection, Ofcom’s broadband and Wi-Fi help pages cover home set-up and your rights, while Thinkbroadband tracks UK network performance and local availability in depth.

How to Fix Bufferbloat

The fix is queue management: making your router handle its waiting data intelligently instead of letting an oversized queue build. The technology is called SQM (Smart Queue Management), and the two algorithms to look for are called fq_codel and CAKE. Instead of one long queue, they keep traffic flowing fairly and drop or mark packets early, so no single download can stall everyone else. The original research project that documented all of this is at bufferbloat.net, still the reference point for the subject.

• Check your router’s settings. Log in to your router and look for QoS, SQM, “traffic prioritisation” or similar. ISP-supplied hubs rarely offer true SQM; some offer a simple QoS toggle that helps a little.
• Set the speeds honestly. SQM works by taking control just before the bottleneck. Set its download and upload limits to about 90–95% of the speeds this test measured — slightly below your real speeds — so the queue forms in your smart router instead of in the dumb buffer upstream.
• Consider a router that supports SQM properly. Many third-party routers, and the open-source OpenWrt firmware, offer CAKE or fq_codel. Put it in charge of the connection (with the ISP hub in modem/bridge mode where supported) and re-run this test.
• Re-test after every change. Change one thing, then test again. A successful fix typically moves a C or D grade to an A or B with no loss of usable speed beyond the deliberate 5–10% set aside for queue control.
• Reduce needless load. Schedule cloud backups and game downloads overnight where the apps allow it. It is not a fix for the router, but it stops the worst collisions while you sort one out.

Troubleshooting: When Results Do Not Add Up

My grade changes every time I run the test

Some variation is normal, especially in the evening. Big swings usually mean the test conditions changed: someone else used the connection, you tested over Wi-Fi, or your line’s peak-time congestion varies. Test wired, at a quiet moment, twice — and compare like with like.

My speeds look low in this test

This test measures speed as a by-product of saturating your line for a few seconds, not as a dedicated benchmark, so figures can read a little below a full speed test. Use our broadband speed test for headline speeds, and this page for the grade.

Testing over Wi-Fi gave me a terrible grade

Wi-Fi adds retransmissions and its own queuing, which look exactly like bufferbloat to any test. Follow our Wi-Fi speed test approach — wired first to judge the line and router, then Wi-Fi room by room to judge the wireless. Do not buy a router to fix a Wi-Fi problem.

The test will not complete

An ad blocker, VPN or strict work or school network can block the measurement endpoints. Whitelist this page, disconnect the VPN, and try again. If it still fails on a home connection, restart the browser and close heavy tabs before re-testing.

Bufferbloat Test – Frequently Asked Questions

What is a good bufferbloat grade?

A or A+ means your latency stays almost flat when your connection is busy, which is what you want. B is acceptable for most households. C or worse means real-time activities like gaming and video calls will visibly suffer whenever the line is loaded, and it is worth applying the SQM fix described above.

Can bufferbloat happen even with fast full fibre?

Yes. Bufferbloat is about queue management, not headline speed. A fast line fills a badly managed buffer just as surely as a slow one — it simply takes a bigger transfer to do it. Gamers on gigabit packages still report it, which is why the grade, not the speed, is the number to watch.

Is bufferbloat my provider’s fault?

Partly and partly not. The queues that hurt you most usually sit in your home router or modem, which is why a router-side SQM setting fixes most cases without your provider changing anything. Congestion deeper in a network can add evening latency too, but fix the home queue first — it is the part under your control, and the part this test measures.

Will a new router definitely fix my grade?

Only if the new router offers SQM (fq_codel or CAKE) and you configure it with speeds set just below your line’s real rates. A faster router with the same unmanaged buffers will not help. After any change, re-run this test — an improvement from C/D to A/B confirms the fix.

Why is my upload grade worse than my download grade?

UK home connections almost always upload far more slowly than they download, so the upload queue saturates sooner and stays saturated longer during backups, video uploads and cloud sync. That is why calls break up when someone sends a large file even on an otherwise healthy line.

How often should I run a bufferbloat test?

When you set up a new router or change queue settings, when you switch provider or package, and whenever calls or games start lagging under load. One test after each change is enough — always wired, always with the house quiet, so results stay comparable.

Test Again After Every Change — and Check the Rest of Your Connection

The routine that actually fixes laggy connections is simple: test, change one thing, test again. Keep this page bookmarked for the re-tests, and build the full picture of your line with the rest of Speed Test UK: headline speeds from the broadband speed test, baseline responsiveness from the ping tester, and latency stability from the jitter test. If you are out of contract and a poor grade has you thinking of switching, compare what is actually available at your address in our guide to UK broadband providers first — a new provider on the same line with the same router habits may change less than a properly configured queue. Start at the Speed Test UK homepage whenever you need to run the numbers again.