/ Technology

My broadband problems brought me down loads

Sad smiley key on computer keyboard

Last month I spent more time speaking to my broadband provider’s helpline than my family. I couldn’t resist telling this to the customer service support staff, and got a sympathetic laugh from the end of the line.

I don’t know what it is about being stuck on the phone trying to sort out technical problems, but I seem to turn into my dad. I can’t control it. It just happens.

Between coming up with awful dad jokes, such as the title for this piece, I was getting more and more frustrated. I’d spent three weeks with speeds under 2Mbps (I should have been getting much faster) and at some points I had no broadband at all.

I snapped. I started doing daily broadband speed tests and became one of those people who just spoke about their internet woes. Working on Which?’s Broadband Speed Guaranteed campaign was an excellent channel for this frustration, but I was no fun at parties.

I’m not alone in my frustration. Our latest broadband satisfaction survey reveals that the biggest providers are letting their customers down. BT, Sky and TalkTalk came bottom of the pile, with smaller providers like John Lewis Broadband, Plusnet and Zen Internet topping the tablet.

Give us broadband speed guaranteed

We also asked customers about their satisfaction with broadband speeds, and most of the providers received a score of three stars or less.

This ups the ante for our Broadband Speed Guaranteed campaign. Not only are we campaigning for providers to improve their service, we want customers to get the speeds they’re promised when they sign up.

At the moment broadband providers are allowed to advertise speeds that only 10% of their customers actually get. That’s why we’re reiterating our call to the advertising watchdogs, The Committee of Advertising Practice (CAP) and The Broadcasting Committee of Advertising Practice (BCAP), to tighten the rules so that advertised broadband speeds more closely match customers’ real experiences. Our executive director Richard Lloyd said today:

‘We’ve told the advertising watchdogs that companies need to be much clearer with their customers about the speeds they can expect. However, three months on, we’re still waiting for them to announce how they’ll ensure adverts only show the speeds most customers actually receive.’

I’m back online

In case you’re wondering, after four callouts my broadband is now back up and running. I was so happy I could have hugged my technician, Winston. (I did.)

Is your broadband service bringing you down loads? We want to hear about it.


Zoe, I see that you give a link above to Which’s speed tester. I have compared Which’s speed tester against several other speed testers on many different days at various times of the day, and it consistently gives a falsely low result that is a small fraction of the speed indicated by other speed testers. I have reported this to Which, but your speed tester is operated by a third party who has done nothing yet to fix this.

I find that the most reliable speed tester is speedtest.net, specifically its Vorboss server in London or its Gigaclear server in Slough.

Whereas speedtest.net shows that my connection is around 880-900Mbps both downstream and upstream and other speed testers give similar results, the Which speed tester always falsely shows that my download speed is only around 100Mbps and my upload speed is only around 200Mbps. These are not small fluctuations but consistent large differences. It is also worth noting that while my ping time to speedtest.net’s UK servers is around 1ms to 2ms, Which’s speed tester shows a ping time of between 21ms and 64ms, which suggests that it is hosted in a high latency environment.

I hope that Which pushes its supplier to sort out these misleading results, particularly as many Which members might consequently falsely accuse their ISPs of slow speeds when in fact it is Which’s speed tester which is at fault.


NFH, I’ve just tried two speed tests, at a time when UK internet use is presumably low – around midnight. On this occasion, the Which? speed test showed a slightly higher speed than speedtest.net – 19.69 versus 18.27 Mbps. But I agree that you just can’t reliably deduce anything from a single speed test, as it could be the server at the other end that’s slow. By the way, do you really mean 880-900 Mbps? That’s nearly Gigabit Ethernet speed – the kind of speed that you’d get transferring files between two PCs sitting side-by-side on your desk linked by a high-speed Ethernet cable.


Hi NFH, thanks for raising this matter about the broadband speed tool on Which.co.uk again.

I can confirm that each speed test has slightly different technology and algorithms. Also, it is very difficult to compare one test to another as you cannot run two speed tests at the same time, and if you run them one after another they may produce different results because network conditions have changed (packets flowing via different routes, different congestion, etc.). The internet is a dynamic environment and all the variables change very frequently. On some connections those fluctuations are very small, so if a user runs the speed test ten times, the variations between the tests will be very low, on some other systems the variations will be higher.

Our test works by trying to maximize the connection throughput by downloading files in parallel. The test at the end then discards the percentage of highest numbers that may be caused by some packets arriving in similar timeframe and therefore artificially showing-up a very short burst of high speed. These millisecond timeframes where speed is very high must be eliminated so we can conclude the maximum throughput accurately.

Since we launched the speed test, we’ve received lots of feedback from users. Some say the speed is high, some say the speed is low. Some feedback was valid and we have used it to make our algorithm better. We examine user reports if they are showing consistently different than expected results by a higher margin of error than 10%.

Of course, I’ll definitely share your comments with the relevant teams here at Which?. 🙂


Clint – Yes, I do mean 880-900Mbps. Here are some recent speed test results at home:


Unfortunately the Which speed tester doesn’t support linking to a previous speed test, so I can’t show a comparison.

I pay £40/month for this quality of broadband.


Andrew – As I said previously, I’m not talking about fluctuations but a consistent large difference between the result given by Which and the results given by other speed testers.

You say that you “examine user reports if they are showing consistently different than expected results by a higher margin of error than 10%“. The difference in my case is consistently 80% to 90%, never as little as 10%. In fact, Which’s speed tester never shows a download speed of more than around 150Mbps or an upload speed of around 200Mbps, whereas other speed testers show 800Mbps to 900Mbps.


That’s quite impressive, NFH, I didn’t realise you could get gigabit Internet speeds at prices not much higher than a typical Virgin cable line. I suspect on this occasion you may have reached the a limitation in Which’s test client software or servers. For example, if the client software uses an interpreted language, such as Javascript or Flash, it may itself become the bottleneck at your sort of internet speeds.


Clint – I understand your point about the software potentially being used, but the high latency suggests that there is some network-related issue. In any case, speedtest.net uses Flash and still shows the full bandwidth.


My download speed is in the range of 7-8 Mbps with both speedtest.net and the Which? speed checker. Both these are scaled up to 100 Mbps, which indicates that this is the intended range of the tests. I would not expect that figures above 100 Mbps would be inaccurate.

We should focus the effort on those – like Zoe in the introduction – who are achieving a slow and/or unreliable service.