Ofcom broadband code of practice changes in pipeline
The Ofcom Broadband Speeds Code of Practice exists in two forms at present, one for consumers and one for business, the key difference is that the consumer guide only talks about download speeds but the business one also adds upload speeds. For the avoidance of doubt there is the minimum guaranteed speed figure, but the guarantee is not as cast iron as the public might assume, simply put if a connection cannot hold the minimum speed consistently and a provider has attempted to resolve any issues you can make the choice to walk from a contract. The danger is if the minimum guarantee is being broken due to physical line issues then walking and switching to another provider may not see things improve substantially (there are scenarios where a different providers hardware may work better with a specific ADSL2+/VDSL2/G.fast line and for VDSL2 this usually means a Broadcom chipset with vectoring and G.INP support).
The code of practices are being updated due to various changes to EU law and one possibility may be that the codes become mandatory, but nothing is certain on that. The changes that are likely to go through revolve around the information to be provided to people and four key things are involved with the main change being the introduction of 'Normally Available Speed'.
- Maximum Speed, the 80th percentile speed at quiet times
- Normally Available Speed, 80th to 20th percentile speeds at busy times
- Minimum Speed, the 10th percentile speed at peak time
- Advertised Speed, what the product is being advertised as to the general populus
The Advertised Speed is current a simple 10% of people with a product should get that speed, but moves are underway to change this to a median or range and 20th to 80th percentile is looking likely. The different with the first three figures in the code is that these should relate to similar lines, so for distance impacted services this means lines of similar length. For fixed speed services such as cable and full fibre which suffer no distance issues then providers have it a bit easier and can simply use the data from those on that specific product.
Defining quiet and peak times will be one of the key things for the Ofcom consultation and may even be something tweaked during the initial implementation window once everyone sees how things actually play out. The process is far from fast so there is lots of time for debate and argue the vairous pro's and con's since the new code is not expected until November after a period of consultation with a soft implementation window to follow and the code is expected to be live in the May to November 2018 window, and even then Ofcom will enter into its monitor phase and changes may take place subsequently.
Building in the difference in performance across the 24 hour day will be useful to help people understand that not all services are created equal even when using the same local loop connection technology, this is something we already do for the main providers every month and Ofcom does itself with an annual report based on a months testing, extended this to the medium to smaller providers is where things become harder, the problem being that quiet times are implicitly when most people are not using their connection, so they is not a good statitiscally useful cohort of tests to draw conclusions from, and for a SME focussed provider peak time might just be 3pm to 5.30pm, but for a provider with a broader customer base it may be 3pm to 11:30pm. The testing systems in place with Ofcom do mean that for the selected lines a lot of data samples are available, and modelling can extrapolate that to a national picture, but always has the risk of the modelling missing nuances, or the deployment of testing hardware and setup notes resulting in resolving wiring issues that mean they are now performance beyond the norm. Other issues arise around new products, since you have not testing to base figures on and there may be services sold to a very small subset of customers (e.g. the old ADSL2+ Annex M is one example).
Our monthly speed test round-ups have shown median (and mean) speeds for providers and where at all possible we identify products for those carrying out tests on our platform (and those using our white label speed test service), starting in July we will make some changes so that ahead of the closing of the consultation there is some public information for providers and others to use, thus we will add figures for the 10th, 20th and 80th percentile for both download and upload on each package. Another change will be that the different technology split figures on https://labs.thinkbroadband.com/local will add this figures to the bar chart that shows the last quarters mean and median speeds. So a fair bit of work to do, since the system is currently set up to record 25th and 75th percentile figures.