Ofcom Start Publishing Raw UK Fixed Line Broadband ISP Speedtest Data

The UK telecoms regulator, Ofcom, has decided to publish some of the raw data that’s used to construct their bi-annual fixed line broadband speeds report (e.g. the August 2013 Report) and have made it available to the public via XLS (MS Excel) and CSV format.

Ofcom’s results are typically based off the installation of special monitoring routers (SamKnows) in around 2,000 homes across the country, which is a small but usually quite accurate sample that allows for detailed testing of connection speed, latency, jitter and packet loss (i.e. the last report performed 736 million separate tests on ISP connections).

But crucially the official bi-annual reports typically only use data collected during May (i.e. this is the report that gets released to the public in August) and November (this will be released next year), while the two additional data releases that have now been made available will instead cover the months of August and February.

Ofcoms Statement

Although these tests are run on a continual basis, up until now Ofcom has only published six-monthly reports, which are based on test results recorded in two measurement periods each year (May and November). Ofcom has decided to issue two additional data releases each year, which are based on tests run in the months of August and February.

These data releases include details of the performance of panellists’ connections on a per-panellist basis, along with comparisons of the performance of a number of popular ISP packages. This will allow consumers and internet service providers to monitor broadband performance on a more regular basis.

The data is presented in a raw, basic format. It does not include UK averages, rural/urban splits or splits by geographic market or technology (as the existing six-monthly reports do).

However Ofcom warns that the new data does not contain the overall UK averages and associated processing and validation, which means that the end-result may not be as accurate as the regulators official report that will continue to be published every 6 months. Never the less this is still useful and interesting information to have.

Some interesting things to note are that almost half (924) of the panellists were made up of 8Mbps capable ADSL1 / ADSLMax and 20Mbps ADSL2+ connections (note: some ADSL1 lines are listed as offering headline speeds of 20Mbps, which is probably due to how the packages are advertised), while cable lines accounted for roughly 300 panellists and the last 700 or so were on 40-80Mbps FTTC with just five users taking BTInfinity’s 160Mbps FTTP package.

But in the real world only 2 million out of around 22 million UK broadband subscribers are currently FTTC based, which suggests that the proportions for a correct reflection of UK performance might be a bit out of whack. Ofcom appears to correct for this in their final report and this is thus another example of why you can’t simply turn the data into direct averages as that would give the UK an average speed – over 24 hours – of a little over 35Mbps (well above August’s 14.7Mbps).

Ofcoms August 2013 Speedtest Data
http://stakeholders.ofcom.org.uk/market-data-research/fixed-line-broadband-perf-updates/