The engine
The measurement code is the same file on BroadbandBoost and Laggy.uk. It talks to the same endpoints that Cloudflare's own speed test uses to send and receive test data (Cloudflare, n.d.), and asks the same network for the connection details it can see, such as your network's name and town. Every logged result carries the engine's name, cf-edge-v1, so results from a different method are never averaged together.
Step by step
| Quick | Standard | Thorough | |
|---|---|---|---|
| Quiet pings | 9 | 15 | 25 |
| Download | 5 s, 4 connections | 8 s, 5 connections | 15 s, 6 connections |
| Upload | 3 s, 3 connections | 6 s, 4 connections | 10 s, 5 connections |
| Busy-line pings | None | Every 0.4 s | Every 0.4 s |
| Takes about | 10 seconds | 20 seconds | 35 seconds |
On a fast line the download adds connections, up to eight, so gigabit lines are not under-read.
How each number is worked out
- Ping. One warm-up request opens the connection, then the quiet pings go one after another. Each is timed from the moment the request starts to the moment the reply starts, with the server's own processing time taken off (Ohlsen & Rodrigues, 2025). Your ping is the middle value.
- Jitter. The average gap between one ping and the next (Cloudflare, n.d.).
- Download and upload. The data moved is counted four times a second. The first quarter of each transfer is left out while the line speeds up. From the rest, the highest and lowest tenth are set aside and the remainder averaged. The trimming is our own method.
- Busy ping. During each transfer, pings start after 0.6 seconds and repeat every 0.4 seconds. The middle value is taken for the download and for the upload separately.
- Busy-line grade. The busier of the two busy pings, minus the quiet ping, gives the rise. The rise sets the grade.
The busy-line grade bands
| Grade | Ping rises by | What it feels like |
|---|---|---|
| A | under 30 ms | Barely any extra lag. Calls and games carry on as normal while others stream. |
| B | 30 to 59 ms | A little extra lag. Most people will not notice outside fast games. |
| C | 60 to 119 ms | Noticeable. Calls can talk over each other and games feel sluggish when the line is busy. |
| D | 120 to 249 ms | Calls and games suffer whenever someone else is downloading or uploading. |
| F | 250 ms or more | Severe. Pages can stall and calls drop while the line is busy. |
These bands are BroadbandBoost's own, shared with every BroadbandSwitch.uk speed test. They are not an Ofcom or industry standard. The idea of measuring ping while the line is loaded follows Cloudflare's test (Ohlsen & Rodrigues, 2025), and the problem it catches is bufferbloat (Gettys & Nichols, 2011).
How "your home on this connection" is worked out
| Room | Rule |
|---|---|
| Films in 4K | One stream for every 25 Mbps of download. Netflix recommends 15 Mbps or more for Ultra HD 4K (Netflix, n.d.); we count 25 per stream to leave room for everything else in the house. Between 15 and 25 Mbps the room shows one 4K stream with little to spare. |
| Online gaming | Smooth with a ping under 30 ms and a grade of A or B. Playable under 60 ms, and noticeable in fast games up to 100 ms, unless the grade is D or F. |
| Video calls | Smooth with 3.8 Mbps up, the most a 1080p group call on Zoom uses (Zoom, n.d.), and a grade of A. Mostly fine with at least 1.2 Mbps up and a grade of A to C. |
| A 100 GB game | 100 GB is 800,000 megabits, divided by your download speed. Under an hour is good, under four hours is fine. |
| Photo backup | 1,000 photos of about 4 MB each is 32,000 megabits, divided by your upload speed. Under 15 minutes is quick, under an hour is steady. |
These are our own rules of thumb, written to be generous. Quick tests have no busy-line grade, so the gaming and calls rooms use ping and upload alone.
Regional and UK comparisons
- Your region comes from the first part of the postcode the test network reports, grouped by postcode area into the twelve UK nations and regions plus the Isle of Man and Channel Islands. A few areas cross a boundary, and each sits in the region where most of its homes are, so the map is approximate at the edges.
- Regional medians use anonymous results from the last 30 days across the BroadbandSwitch.uk speed tests. A region shows a figure once it has at least 20 tests. No individual result is ever shown.
- "Faster than X in 10 nearby" places your download within your region's spread, using the 10th, 25th, 50th, 75th and 90th percentiles and drawing a straight line between them. It is an estimate, rounded to the nearest tenth.
- The UK median is the middle download speed for the last full calendar month, shown once it rests on at least 20 tests.
The figures come from the shared results service on Laggy.uk, which publishes regional medians and monthly UK figures as open JSON.
What is stored
If result logging is on, one anonymous row per test: speeds, pings, jitter, the busy-line grade, the test length, your Wi-Fi or cable answer, the network name and number, town, region and postcode district, the test location, whether the connection uses IPv6, the test's length in time and data, device and browser family, the time and the engine name. Never your IP address: the results service keeps only a one-way code made from it with a secret key. Rows are deleted after 13 months. The privacy notice has the detail.
What a browser test can and cannot tell you
- It measures the whole path from your device to a nearby Cloudflare server, including your Wi-Fi, your router and your device. That is what you experience, but it is not the same as your provider's line test.
- A test server close to you gives the best case. Services hosted further away will see more delay.
- One test is a snapshot. Ofcom suggests several tests over several days at different times (Ofcom, 2024).
See it on your own line
Run BroadbandBoost to get your download, upload, ping and busy-line grade, then compare deals at your postcode if it falls short.
Sources
- Cloudflare. (n.d.). cloudflare/speedtest: Component to measure the quality of a client's Internet connection [Computer software]. GitHub. https://github.com/cloudflare/speedtest
- Gettys, J., & Nichols, K. (2011). Bufferbloat: Dark buffers in the Internet. Queue, 9(11). https://queue.acm.org/detail.cfm?id=2071893
- Netflix. (n.d.). Netflix-recommended internet speeds. Netflix Help Center. Retrieved November 16, 2026, from https://help.netflix.com/en/node/306
- Ofcom. (2024, September 24). Practical tips for improving your broadband speed. https://www.ofcom.org.uk/phones-and-broadband/coverage-and-speeds/broadband-speeds
- Ohlsen, L. Y., & Rodrigues, C. (2025, October 27). How does Cloudflare's Speed Test really work? The Cloudflare Blog. https://blog.cloudflare.com/how-does-cloudflares-speed-test-really-work/
- Zoom. (n.d.). Calculating bandwidth usage for Zoom meetings and phone. Zoom Library. Retrieved November 16, 2026, from https://library.zoom.com/admin-corner/network-management/quality-of-service-and-network-best-practices-explainer/calculating-bandwidth-usage-for-zoom-meetings-and-phone
Cite this page
Martin-Smith, A. J. (2026, November 16). How BroadbandBoost measures. BroadbandBoost. https://broadbandboost.co.uk/methodology.html