Ofcom has published a biannual update for their latest dataset on fixed broadband speeds, which among other things found that the average (mean) actual download speed of UK residential ISPs has increased from 64Mbps (14Mbps upload) in their May 2020 report (here) to 71.8Mbps (14.2Mbps) now.
The regulator’s study is based off data gathered via custom modified routers from SamKnows, which are installed in a few thousand homes. This method of testing is extremely accurate (i.e. tests avoid WiFi and are only done during idle periods), but the limited sample size does tend to limit their focus to only the most common broadband technologies, packages and ISPs.
The latest data is also fairly raw and not as detailed or extensive as their main annual report, which normally comes out in the Spring. As such we’re only going to post a very short summary of the key figures. All of the speeds stated below are downstream, unless otherwise stated.
Advertisement
The regulator has also taken an updated look at the differences in broadband speed between urban and rural areas on copper ADSL and hybrid fibre FTTC (VDSL2) lines. As ADSL and FTTC copper lines tend to be shorter in urban areas than in rural ones, urban lines tend to perform a bit better than those in rural areas (lower signal degradation over distance).
Advertisement
Finally, Ofcom’s short update also gives us a look at the differences in average latency times (lower figures are better) between FTTC broadband packages on UK ISPs Sky Broadband, BT, EE, Plusnet and TalkTalk (see below). As you’d expect they all have broadly similar performance characteristics and a difference of 1-2ms (milliseconds) here or there isn’t really worth worrying about.
Just for good measure they also seem to have included the “full fibre” FTTP variant of BT’s 67Mbps package, which naturally shows a much more obvious performance benefit over FTTC. We have seen other FTTP networks delivering even better results than this.
I have slower download than the average, 65mbps, but faster upload, 18.7mbps. Stuck on like a 250m long line or something like that. Would like faster but no fttp in sight or anything which is a big shame.
That is very strange and does t make sense.
With an upload of 18.5mb/s I would expect 68-72mb/s download.
A builder – if only life were that simple! The relationship between download and upload speed is not an exact formula unfortunately.
CarlosTheGuru
Actually it should be.
It simply depends on the electrical properties of the copper pair.
When the ratio is off it usually indicates a fault.
For instance we had 18/18 service on FTTC and this was down to a dud port in the DLSAM.
With a relatively short, by FTTC standards, line like this I would expect to get pretty good performance.
I had a full 20Mb upload sync on my line but only got around 65Mb download too. It was basically down to crosstalk in the very busy cabinet rather than a “fault”.
A Builder – ok, I take your point that in an absolutely ideal world (where cute furry animals don’t die at the hands of nasty people) then downstream to upstream bandwidth is perfectly in the proportion 4 to 1 on FTTC.
If you take a look at BT’s Broadband Availability Checker:
https://www.broadbandchecker.btwholesale.com/#/ADSL
you’ll see that BT (Openreach) give every line an estimated range for both the down- and upload bandwidths. There’s also a Handback Threshold which they will allow the line speed to drop to before declaring it to be in a fault condition.
You can expect what you like, but don’t expect Openreach to come and investigate why your downstream bandwidth isn’t exactly 4 times your upstream (just because it should be).
CarlosTheGuru
I agree with you that 65mb/s isn’t a disaster on OR FTTC anyway.
I would be a waste of resources to send engineers out to chase down the smalll Improvment.
I would hazard a guess that the issue is at the consumer end anyway and that a router/modem/firmware update is the necessary.
XD, I’m on a 40Mbps MAX line – I get 35 Mbps down and a lovely 5 Mbps up! Would love to have 75Mbps.
“Actually it should be.
It simply depends on the electrical properties of the copper pair.”
That couldn’t be further from the truth. I’m understating that.
You have to take in to account DPBO, UPBO, PSD masks, the CAL value of the DSLAM, modem chipset, DSLAM chipset, DLM etc, etc, etc.
Power levels are very different for cabinets close to the exchange compared to cabinets that are far from the exchange to minimize interference with ADSL services.
That’s something that makes every line and every cabinet different.
There is no correlation/formula between downstream/upstream.
You can just as easily have 80/15 as you can have 50/20.
That’s before you start taking in to consideration all the variables an individual line can have.
There’s noise that can effect particular frequencies/tones.
There are rogue modems with settings that can cripple the upstream of other lines within the same bundle.
There are literally dozens and dozens of xDSL chipsets that all perform different. Some sync lower on the Downstream and higher on the upstream and vice versa.
Then there’s the ECI/Huawei difference.
G.INP and a 3dB SNRM target can be worth as much as 15Mb+ compared to a line with interleaving.
On my ECI cabinet my line would sync at 30/5.
On my Huawei cabinet my line syncs at 44/7.
So even 2 neighbours with the exact same line length, on the exact same DSLAM, even without all the variables i mentioned above and if there was a correlation between downstream/upstream (there isn’t), they could be 65/20 and 80/20 just from DLM settings alone.
71mbps is the average? For where? Our whole village gets around 4-7mbps and bt have said “no plans to upgrade you”. Were looking at the community grants and stuff now , but it’s taken years to build interest.
What do you mean for where, do you know how average works? If you have bunch of 100-300 or 1000mbs offers in cities that rack ups average.
Getting left even further behind, while others get faster and faster, and then still complain!
If everyone with poor speeds requested a box from SamKnows it would bring the average down. Might make regulator and ISPs take note!
FYI you can see all sorts of info about your own line from one of these boxes.It can also help you diagnose whether it is WiFi or line causing the issue (as Mark said they bypass the WiFi and plug directly into your router)
One 900Mbps user, averaged with a dozen 4Mbps users, makes an average of 72.9Mbps.
Hence it only takes a few 900M users to wipe out large numbers of people with poor speeds. It makes OFCOM look good, because of the “progress” it shows.
More interesting would be the median (50th percentile), and say the 5th percentile – that is, the speed that 95% of users exceed, and 5% of users cannot achieve.
If 100’s of long line adsl people got a sk box it would not change result.
The average is NOT a simple mean of all the boxes, there is modellling applied.
The stats are pretty meaningless, it’s like comparing apples and pears, with all the different technologies. DSL is not the same as fibre.
The problem with these stats is that it clouds (pun intended!) the fact that the “small” percentage of connections on less than 5mbps is actually a large number. A high average like this makes it easier to ignore the number of lower speed connections.
The promise of 10mbps (let alone the joke of gigabit by the government) was, and still is, virtually impossible to achieve with the current USO as far as I can see.
Eventually they’ll just invent a stat that deletes these speeds from the calculation.
I work in ISP tech support and we get loads of central London customers with less than 5mbps – if the government can’t sort those out, what hope does the rest of the country have?
And that I totally agree with.
From my experience of setting up site offices in Central London connectivity was terrible.
If is getting better really fast with GNetwork, Hyper and Community doing a great job. Looks to me like some of the streets in RBKC will get OR FTTP soon: saw the fibre teams out last week.
Er… I wish.. stuck on eci cab…
Yeah! I’m below average again, feels great 🙂
Plusnet performing better then BT and EE now that’s funny. I’ll stick to IDNET though.
As they say “there are three kinds of falsehoods, lies, damned lies and statistics” and the headline of this report clearly is the former. If you look at the underlying data it does not appear to represent the current UK BB landscape and whilst the file is a good representation of individual technology performance, UK wide conclusions and mixing of technologies should not be concluded from it.
In simple terms HFC and FTP technology will perform (connect not ISP) at, near or in VMs case slightly higher than the purchased service. Averages of these simply swamp the variability and low performance DSL products.
In addition if people discarded price/ISP preference and went for the fasted speed available then we would be looking at over 360Mbps as an average. Meaningless to those struggling on broadband below the USO.
What should be taken from the sample is an indication to the real average performance of a given technology, especially ADSL and FTTC where these AVERAGES clearly show Ofcom that high percentages on these products they are experiencing well below expectation and that they are overstating our broadband status, customers are being over charged and there appears to be little Ofcom are doing.
Technology Product Expectation Average Download
ADSL1 8-10Mbps 3.32
ADSL2+ 10-20Mbps 10.03
Cable Ultrafast 223.58
Fixed Wireless Superfast 32.69
FTTC Higher 55-76Mbps 57.13
FTTC lower 38Mbps 30.35
FTTP Full Fibre 171.93
Just goes to show how completely living out of reality OFCOM is if they believe these stats. There is a huge chasm of a difference between “ISP Speeds Rise to 71.8Mbps” and what users actually get.
You don’t like how maths works so you decide the problem must be with ofcom?
Faster fibre 41 quid a month with talk talk,max 15 Mbps ,crap
Most Humans have an above average number of arms and legs. /shrug
Only for mean. Most humans have exactly the average number of legs when median and mode are used.
Actually the UK national average download speed is 46.23Mbps as of November 2020! According to Broadband UK https://www.broadband.co.uk/broadband-speed-test/
The reason you see 71.8Mbps is because of an artificial increase due to Virgin Media, G.Fast and FTTP. Obviously if you have customers choosing 300-1000Mbps across the UK, this will create the illusion that the averages are higher. It saturates the statistic.
But if you exclude FTTP, G.Fast and Virgin Media and solely take a statistical measurement using FTTC and ADSL only, that average drops to 46.23Mbps as quite rightly shown according to the bar chart.
Simply put, most people do not have Fibre cabinets within 300-400 meters proximity from their property when you include rural areas and towns, so they are not going to see speeds at 70+Mbps. If this was the case, the country wouldn’t be so desperately needing an FTTP upgrade.
It’s not an “artificial increase”, this is just how averages work.
Yes, that means that the national average speed isn’t very useful for working out what speed you can expect at your location. It doesn’t mean that excluding a bunch of services from the average gives you some kind of “actual average”. That will just give you the average over the services you included.
You need to exclude most of the other services because they are currently a niche compared to the 95% of the country who mostly access their internet using FTTC as a service.
That’s the problem with this study. That’s not average, average is when the average user gets around 71.8Mbps across the country.
The 46.23Mbps National Average by Broadband UK is a more genuine average stat as it excludes FTTP as a service. When we gather a statistic, it is important to know what most people get with FTTC. It is completely irrelevant to include average speed stats into the equation from services that most people don’t have access to. Yes, it will improve the UK’s reputation in the Worldwide Broadband Speed League table, but it is not beneficial for the individual average user.
In future when the UK will achieve 100% Gigabit coverage, the average speed will no doubt shoot up to over 100Mbps. Only then the study will be more accurate.
They can’t exclude FTTP, it uses the same address ranges as normal, and I doubt they exclude Virgin Media given it’s available to the majority of the country. FTTP is available to nearly 1 in 5.
More likely it’s the impact of people using wireless to test.
Still, yes, clearly a comparison site (that offers products that aren’t available) relying on people testing, something most do for a reason and will usually be done over wireless, is definitely far more reliable than SamKnows.
This average figure isn’t any less genuine for including non-FTTC services, and you can’t make it more genuine by excluding them.
That said, you’re right that it’s not very useful if you’re wondering what sort of speed you’ll get in a particular area… but an FTTC-only figure isn’t very useful for that either, since FTTC’s speed depends strongly on location and line condition. This sort of national average would work a lot better for services that aren’t distance-dependant.
If anything, Virgin Media doesn’t distort the statistics enough. With a typical OpenReach connection most people will simply have the maximum speed their line supports, whereas most VirginMedia users have a line that supports 1Gbps but most would have chosen the cheapest 100Mbps package.
Mike, with an Openreach FTTP or any FTTP connection there isn’t a question on what line can support the maximum speeds. If you were to order an 80/20 Mbps FTTP package you are pretty much guaranteed to get that connection speed.
Therefore with FTTP, the average UK speed will in fact increase way higher even if most people chose a 100 Mbps package instead of a Gigabit package.
FTTC has a major drawback that DLM is the determinant on what kind of speeds you’ll get and will settle at that speed depending on stability.
Now I am lucky that I get 80/20 Mbps on FTTC using TalkTalk but that’s only because my cabinet was upgraded within 320 meters from my property. But even now 10 months in, I am seeing some minor issues. At the start of February I was syncing at 80 Mbps with a 6dB SNR profile. As more people seemed to have joined, particularly at lock-down, my noise margins are now at 2.30dB. This is a very low SNR. Fortunately my connection is still maintaining sync, but who knows for how long! If crosstalk becomes more intensive I probably won’t sync at 80 Mbps anymore.
My speed did drop at one point temporarily to 69 Mbps in June following a firmware update that triggered DLM interleave. I had to wait again around 2 weeks for the connection to gradually climb back up to 80 Mbps.
With FTTP we will never have this concern. Indeed with Virgin Media alone it won’t distort the statistic that much. The 71.8 Mbps average national figure is definitely never going to happen from FTTC alone only. It’s impossible, unless Openreach started upgrading new street cabinets to make them closer to absolutely everyone, which is not going to happen.
That 46.23 Mbps average is not because most are using a wireless connection. Even if they all used wired it’s not going to be higher unless the router sync speed was also high, which indeed is not the case for everyone. E.g I have always used wired connection on ADSL previously on EO Line but 12 Mbps was the max stable sync I could achieve.
“You need to exclude most of the other services because they are currently a niche compared to the 95% of the country who mostly access their internet using FTTC as a service.”
A niche?
95% might have access to FTTC.
That doesn’t mean 95% use it.
Virgin media is available to over half of UK homes. How is that niche? Just because you don’t have it?
OpenReach FTTP is available to over 3.5m homes and G.Fast to a further 2.3 million homes. That’s not niche either.
You don’t get an average by excluding over half of the country.
You want to exclude all of the faster technologies that are available to over 60% of UK homes.
100% of homes have access to dialup.
99%+ have access to satellite broadband.
More can access dialup than anything so let’s take the average from that.
Sums up your logic nicely.
@John: Look at the Subscriber List Top 10 here on ISPreview. Virgin Media has only 5.3 million subscribers despite being available to just over half the country. While all the rest of the ISPs add up to over 20 million subscribers. Most of these 20+ million subscribers are FTTC users.
3.5m Openreach FTTP is a niche in ratio to the whole UK population and the same with G.Fast. They are still able to make a contribution to the 71.8Mbps average rise. But when you exclude them and count FTTC figures only, that number drops to 46.23 Mbps.
The other thing is that Virgin Media is extremely expensive at the end of the contract for their lowest package at 108Mbps it costs £51 a month. This means that at the end of the contract most people switch back to an FTTC package that is in the low 20s. While FTTC customers usually end up switching to another FTTC provider at the end of the contract.
Bringing up the dial-up argument is just a logical fallacy. Hardly anyone uses dial-up at this stage. Satellite broadband may be available to 99% of the country, but most premises particularly private estate buildings won’t allow you to install a satellite dish, so it is as if it is not available.
You need to take the average from the most popular choice that consumers choose from what is available. At the moment FTTP is only available at 18.5% coverage. When it becomes 100% coverage/ and or when Gigabit coverage will go from 36% to 100% then we can make a more accurate assessment.
It’s all comes from where ever based at the time and the power outage that,s powering our transformers that gives trade markets a boost then when these lose power signals through power from passing planetary objects you cannot comprehend,signals go weaker, and billions go missing ️️️⚖️⚖️⚖️G
What areas was this tested in ?
Mine is getting worse, 35mpbs down and its dropped down to 3.5mbps up/ previously we had between 9 and 10mbps up, we are with Vodafone in Dorset, I’ve also a neighbour who tested the same speeds 2 doors away from me, again Vodafone. Their 3-5 days fix time has passed. Nothing has changed. Sorry I know its off topic, just frustrated.
It’s a national average.
But as a post above says using mean average instead of median means that the majority of people will get less than the ‘average”