Broadband News

Better broadband speed info on the way for the public

The Ofcom Broadband Speeds Code of Practice has existed for some time now and a new revised 2019 version now exists that takes into account the changes around broadband advertising, i.e. the shift from top 10% speed in adverts to the median peak time speed. The code does remain voluntary, but with most of the well known household names signed up those providers who do not join may find people asking more questions as to why.

The paperwork around the new Code of Practice is on the Ofcom site and the list of providers signed up for the 2019 version are:

  • BT
  • Daisy
  • EE
  • KCOM (Hull Area)
  • Lothian Broadband
  • Plusnet
  • Sky
  • TalkTalk
  • Virgin Media
  • XLN

The new code of practice comes into effect on 1st March 2019 but a 12 month implementation period will exist with compliance only being tested in 2020, therefore while some providers who have completed work to offer the new detail may update their sales and post sale information very soon for some other signatories it may be some months before changes are seen.

More realistic speed estimates at the point of sale. Speed estimates provided to customers at point of sale should reflect the speeds that they are likely to experience at peak times. This speed will take account of the fall in speeds that occur during peak-time network congestion, and is more reflective of the speed a customer will receive at the point in the day that they are most likely to be using their broadband service. Peak times will be measured as 8-10pm for residential services and 12-2pm for business services;

Always providing a minimum guaranteed speed and the right to exit connected to this speed at the point of sale. This will ensure that customers are aware of their right to exit their contract if speeds fall below a minimum level;

Strengthening customers’ rights and extending the right to exit to bundled products. The right to exit will now apply to bundled products, such as landline services on the same line, or pay-TV services purchased at the same time as the broadband service. A new 30-calendar day limit will apply to the time providers have to improve speeds before they must offer the right to exit to customers, and providers will be required to make information about the right to exit in after-sale information more prominent and to link it more clearly to the minimum guaranteed speed so that customers understand what triggers this process.

Ensuring all customers benefit from the codes, regardless of their broadband technology. The existing codes only apply in full to broadband services provided over certain networks such as copper and part-fibre. By moving to an approach that measures customer speeds at peak time, it is possible for the codes to apply in full to all access technologies. Under the new codes cable and fibre to the premise (FTTP) providers will also be required to provide detailed speed information to customers at the point of sale, including the normally available estimate and a minimum guaranteed download speed, which triggers the right to exit.

Key parts of Ofcom Broadband Speed Code of Practice

Embracing full fibre and cable services is a significant change and with full fibre services growing rapidly now it was important to bring them on board and maybe reign in some the mystique attached to full fibre services.

Another change is that upload speeds will need to be shown alongside the download speed estimates at the point of sale and in customer portals for existing customers. This is less of a concern than a couple of years ago when the 40/2 FTTC service was sold a lot more, but the discounts at the wholesale level for the 40/10 service meant most providers have stopped selling it and have also upgraded all existing customers.

The new 30 day limit for providers to resolve speed problems is welcome as it will focus a provider on getting a resolution in place rather than what can happen now of the public stuck in a never ending loop of testing. The right to exit that the public can exercise also does allow providers to do things like offer a discount instead or an upgrade to a faster service, which provides a useful remedy for those with limited provider choice e.g. if your broadband options are ADSL2+ at just 2 to 3 Mbps or a FTTP operator at much higher speeds but they do have some congestion issues you are very unlikely to self downgrade to the ADSL2+ service.

The 8pm to 10pm peak time period for residential services is of no surprise but a business peak of midday to 2pm is actually very revealing about how business use of broadband, the suggestion seems to be that staff doing a bit of browsing while eating lunch at their desks is pushing bandwidth use up (the business window is based on analysis of data usage) and suggest that in terms of utilisation of broadband connections many firms are not pushing the boundaries of the technology they are using.

The way that providers are meant to determine the minimum guaranteed speed (often called MGAL) and the other speed range figures two worked examples taken from an Ofcom annex provide the clearest way to present this.

Examples of how to apply the methodology to obtain download speed estimates for a 76Mbit/s FTTC package

A representative sample of the product’s customers is tested. For each customer, their mean peak time speed is compared to their current sync speed.

For the whole panel, the mean difference between peak time and sync speeds is calculated to be a 20% reduction. The difference between quiet hour and sync speeds is calculated to be a 10% reduction.

A new customer provides their address from which the relevant group of similar lines is identified. The network infrastructure provider tells the ISP that the 80th, 20th and 10th percentiles of this group’s sync speed distribution are 60Mbit/s, 55Mbit/s and 40Mbit/s respectively.

To derive their speed estimates, the congestion measure is applied to these speeds as follows:

  • Normally available speed (20th to 80th percentiles, adjusted for peak):
    55Mbit/s to 60Mbit/s, reduced by 20% = 44Mbit/s to 48Mbit/s
  • Minimum guaranteed speed (10th percentile, adjusted for peak)
    40Mbit/s, reduced by 20% = 32Mbit/s
  • Maximum speed (80th percentile, adjusted for quiet hour)
    60Mbps, reduced by 10% = 54Mbit/s

200Mbit/s cable package and same applies for full fibre services

A representative sample of a product’s customers is tested nationally. Two data distributions are created: one with each panellist’s mean peak time speed and one with each panellist’s mean quiet hour speed.

To derive the new customer’s speed estimates, this data is used as follows:

  • Normally available speed
    The 20th and 80th percentiles of peak speeds are derived e.g. 160Mbit/s to 185Mbps
  • Minimum guaranteed speed
    At least 50% of the 200Mbit/s product speed e.g. 100Mbit/s
  • Maximum speed
    The 80th percentile of quiet hour speeds is derived e.g. 205Mbit/s

The minimum guaranteed speed for the ADSL, ADSL2+ and VDSL2 (FTTC) services is now not an access line speed (sync speed) but a mixture of both sync speed and throughput, which means we may see some more variation between the different providers, though the VDSL2 operator with the largest likely peak time variation is not signed up to the code of practice i.e. Vodafone.

The measurements for determining how peak time congestion is impacting download speeds is based on measurements at the broadband router, thus negating Wi-Fi and device issues and providers are not allowed to use on-net testing. The testing is mandated to be involve three concurrent downloads of a HTTP file and must run for 5 seconds after the TCP ramp up and testing should not run when other traffic is detected at the CPE, so if you are someone who is always streaming TV and movies in the evenings you will not be feeding into the congestion data set even if the provider has spent the money to add the testing software to the router firmware.

The minimum guaranteed speed does not guarantee that all downloads will be above that threshold, since it is impossible to provider any such blanket guarantee and with the 30 day period to fix a single evening of abysmal performance will not trigger the right to exit because once you've reported the problem after a day or two of testing it will be shown that things have improved. Where this will help is when slow speeds happen night after night and impact multiple sites and services.

One point worth making is that with the guarantee using three concurrent downloads this does not mimic the standard behaviour of streamed content, so the random buffering some see but speed testing on some sites suggests everything is performing perfectly will not be reflected - and this is why our speed test does the single download test as well as following up with the multiple download test.

Not really surprising but Virgin Media during the consultation phase did not support the peak time period, but preferred a 24 hour average to be used which of course would mask peak time congestion issues and while people may use their broadband 24/7 the vast majority of people are still home and using their broadband the most between 6pm and midnight.


How would a customer establish that their peak time download speed falls short of what was promised when the order was placed?

  • Michael_Chare
  • about 1 year ago

Most likely people would spot problems via the usual speed tests, and then when contacting the ISP to complain about poor speeds further checks would be done e.g. kicking off a CPE based speed check manually.

The other way is if the sync speed falls below the minimum guaranteed figure, at which point the usual household wiring and other support issues providers should all know about kick in. Of course sync speed should not be impacted by peak time issues.

  • andrew
  • thinkbroadband staff
  • about 1 year ago

But could for example an ISP get away with insisting that a speed test that uses multiple streams be used? The location of the speed tester could also be relevant. I would have thought that Ofcom should have stipulated how speed tests are run for the purpose of establishing non compliance.

  • Michael_Chare
  • about 1 year ago

The testing to determine peak time drop must be three HTTP concurrent downloads, and run for 5 seconds after initial ramp

Files must be hosted off-net

So yes Ofcom is specifying how the testing should be done - all buried away in the Ofcom paperwork

  • andrew
  • thinkbroadband staff
  • about 1 year ago

Thank you. Certainly well buried.

  • Michael_Chare
  • about 1 year ago

if you want to compare broadband deals and plan you can visit

  • garima7
  • about 1 year ago

Post a comment

Login Register