UK fixed-line broadband performance, May 2014 - The performance of fixed-line broadband delivered to UK residential consumers

Published 03|10|14

Overview of UK fixed broadband speeds

1.1 The average UK residential fixed broadband speed reached 18.7Mbit/s in May 2014

Our research found that the average actual speed for UK residential fixed-line broadband connections was 18.7Mbit/s in May 2014 (Figure 1.1). The average actual speed of superfast fixed broadband connections (which in this report are defined as being those with headline speeds of 30Mbit/s or more) was found to be 47.0Mbit/s, more than five times the average actual speed of those connections with headline speeds above ‘up to’ 10Mbit/s and less than ‘up to’ 30Mbit/s (9.0Mbit/s). The average actual speed of connections above ‘up to’ 2Mbit/s and up to and including 10Mbit/s was 3.3Mbit/s.

Figure 1.1       Average UK broadband speed: May 2014

Source: SamKnows measurement data for all panel members with a connection in May 2014
Panel Base: 956
Notes: (1) Data have been weighted by ISP package and LLU/non-LLU connections, rural/urban, geographic market classification and distance from exchange to ensure that they are representative of UK residential broadband consumers as a whole; (2) As sufficient sample sizes were not available for consumers on packages of ‘up to’ 2Mbit/s or less, data collected for these packages in April 2009 has been factored in, in proportion to share of all connections in May 2014; (3) Data collected from multi-thread download speed tests; (4) The above ‘up to’ 10Mbit/s and less than ‘up to’ 30Mbit/s category includes ADSL2+ connections which are not marketed using a connection speed.

1.2 Average broadband speeds increased by 0.9Mbit/s in the six months to May 2014

Residential fixed broadband speeds continued to increase in the six months leading up to May 2014 when the average actual speed of these connections (18.7Mbit/s), represented an increase of 0.9Mbit/s (5%) on November 2013 (Figure 1.2). This was a marked slowing in the rate of increase, as between May 2013 and November 2013 the average speed increased by 3.1Mbit/s (21%).

Over the last few years, increases in average residential fixed broadband services have mainly been driven by increasing take-up of superfast broadband services and average speeds among these connections. However, the average speed of superfast connections was unchanged at 47.0Mbit/s in the six months to May 2014 (see section 1.4 for more details), and as a result the increase in average speeds across all connections was lower than in previous periods. Average actual download speeds also remained stable for connections with a headline speed above ‘up to’ 2Mbit/s and up to and including 10Mbit/s in the six months to May 2014, at 3.3Mbit/s, while the speed for above ‘up to’ 10Mbit/s and less than ‘up to’ 30Mbit/s increased by 0.6Mbit/s (8%) over the same period.

Sufficient sample sizes were not available among our panel for connections with headline speeds of ‘up to’ 2Mbit/s and less, because of the low market share of these connections, so the performance of these connections is not analysed in detail. They are however included in Figure 1.2 so that average actual speeds are reflected across all connections.

Figure 1.2       Average actual broadband speeds: November 2008 to May 2014

Source: SamKnows measurement data for all panel members
Notes: (1) Data have been weighted by ISP package and LLU/non-LLU connections, rural/urban, geographic market classification and distance from exchange to ensure that they are representative of UK residential broadband consumers as a whole; (2) As sufficient sample sizes were not available for consumers on packages of ‘up to’ 2Mbit/s or less, data collected for these packages in April 2009 has been factored in, in proportion to share of all connections in May 2014; (3) Data collected from single-thread download speed tests prior to November/December 2010 and multi-thread download speed tests for November/December 2010 onwards; (4) The above ‘up to’ 10Mbit/s and less than ‘up to’ 30Mbit/s category includes ADSL2+ connections which are not marketed using a connection speed.

1.3 Twenty-eight per cent of UK residential broadband connections had a headline speed of ‘up to’ 30Mbit/s or higher in May 2014

We consider the increased take-up of superfast connections to be a major driver in the increase of actual speeds across all connections. The proportion of connections of residential fixed broadband that had a headline speed of ‘up to’ 30Mbit/s and higher was 28% in May 2014 an increase of four percentage points from November 2013. The proportion of connections with headline speeds over ‘up to’ 10Mbit/s and less than ‘up to’ 30Mbit/s fell by three percentage points to 61% in the six months to November 2013, consistent with the decrease for these connections between May 2013 and November 2013, as consumers upgrade to superfast services.

The proportion of connections that were ‘up to’ and including 8Mbit/s to 10Mbit/s also fell, down by two percentage points in May 2014 to 10%, while connections with a headline speed of less than ‘up to’ 8Mbit/s accounted for less than 1% of connections.

 

Figure 1.3       UK residential broadband connections, by headline speed

Source: Ofcom, based on data provided by the UK’s largest ISPs by retail market share (representing over 90% of the total market)
Note: The above ‘up to’ 10Mbit/s and less than ‘up to’ 30Mbit/s category includes ADSL2+ connections which are not marketed using a connection speed.

1.4 The average speed of cable broadband services overtook that of fibre connections in the six months to May 2014

The average actual download speed of fibre broadband (FTTx) connections was 42.0Mbit/s in May 2014, down from 42.9Mbit/s in November 2013, a reduction in speed of 2%, continuing the trend of falling average fibre broadband speeds seen in the six months to November 2013.

In the UK, most residential fibre broadband connections (around 99% of the total) are ‘up to’ 38Mbit/s or ‘up to’ 76Mbit/s fibre-to-the-cabinet (FTTC) services. In the six months to May 2014 the proportion of these connections which were ‘up to’ 38Mbit/s increased slightly and this, coupled with small decrease in the average speeds recorded for both ‘up to’ 38Mbit/s and ‘up to’ 76Mbit/s FTTC connections, resulted in the observed decline in the average speed for fibre broadband connections.

For all other technologies average actual download speeds increased, with the average speed of ADSL connections increasing by 10% to 7.4Mbit/s (possibly a result of infrastructure improvements related to fibre rollout) and cable speeds increasing by 3.2Mbit/s to 43.3Mbit/s. The increase in the average speed of cable connections is partly due to Virgin Media’s launch of a new cable speed upgrade programme in February 2014. However, the increase in average actual cable broadband speeds slowed in the six months to May 2014 (from 15% in the six months to November 2013 to 8% in the six months to May 2014), possibly due to the current upgrades being opt-in (rather than automatic as was the case with Virgin Media’s last upgrade programme).

Despite this slowing, the increase in the average speed recorded for cable connections during the period (and the fall in average fibre broadband speeds outlined previously) was sufficient to result in the average actual cable broadband speeds being higher than that of fibre broadband services for the first time since we have collected comparable data for the two.

Figure 1.4       Average broadband speeds for fixed broadband connections, all connections including ‘up to’ 2Mbit/s and less, by technology

Source: SamKnows measurement data for all panel members with a connection in May 2014
Panel Base: 956
Notes: (1) Data have been weighted by ISP package and LLU/non-LLU connections, rural/urban, geographic market classification and distance from exchange to ensure that they are representative of UK residential broadband consumers as a whole; (2) Data collected from multi-thread download speed tests.

1.5 The gap between rural and urban speeds appears to have narrowed slightly

Average speeds in urban areas tend to exceed those in suburban and rural areas due to the higher availability of cable and fibre services and because average line lengths are longer in rural areas due to population density being lower. Longer line lengths between local exchanges and customers’ premises result in signal degradation, and actual ADSL speeds being lower, while longer line lengths between the street cabinet and customer premises result in lower speeds over FTTC connections.

The higher availability of cable and fibre in urban areas is due to the fact that initial network rollout tends to be concentrated in areas where population density is higher, resulting in a higher number of premises being covered for a given network investment and giving a larger potential customer base.

In the six months to November 2013, average actual download speeds in urban areas increased by 21%, while suburban speeds grew by 22% and average rural speeds grew by 15% (although this increase was not statistically significant). However, our analysis suggests that in the six months to May 2014 average rural speeds experienced the highest growth (up from 11.3Mbit/s to 13.6Mbit/s, a 20% increase), while in both urban and suburban areas average speeds increased by just 5% (to 33.4Mbit/s and 22.9Mbit/s respectively).

None of the recorded increases in average speeds in urban, suburban and rural areas or the reduction in the gap between average urban and rural areas were statistically significant to the 95% level of confidence used in this report in the six months to May 2014. However, the data suggests that the gap between speeds in rural and urban areas may be starting to narrow, as we would expect to happen as the rural availability and take-up of fibre broadband services increases. This is supported by the increases in speeds recorded in rural areas in the six months to May 2014 being concentrated among a small proportion of rural panellists.

Figure 1.5       Average download speeds for fixed broadband connections in urban, suburban and rural areas: May 2011 to May 2014

Source: SamKnows
Panel Base: 2011; Urban 999, Suburban 382, Rural 323; 2012 Urban 1099, Suburban 391, Rural 294; and 2013 May Urban 1362, Suburban 448, Rural 365; 2013 November Urban 746, Suburban 292; Rural 271; 2014 May Urban 1280, Suburban 503, Rural 370
Notes: The UK average above will not match shown elsewhere in this report as different weightings have been used to enable us to provide comparisons by urbanity. Further statistical methodology is provided in Annex 3.

1.6 Rural areas are affected more by contention at peak-times than urban areas when using both ADSL and fibre technologies

The low availability of cable broadband services to rural areas means that ADSL and fibre broadband services are more prevalent in rural areas. Our May 2014 research shows that rural ADSL and fibre connections provided lower average download speeds than those in urban areas. Rural ADSL connections had an average actual download speed of 5.1Mbit/s over the 24-hour period, compared to 9.7Mbit/s for urban connections. For fibre connections, urban consumers achieved a 24-hour average actual download speed of 49.8Mbit/s compared to 40.8Mbit/s for rural consumers.

Contention affects rural consumers more on average than urban consumers: rural ADSL connections received an average of 85% of their maximum speed during the 8pm to 10pm weekday peak period, compared to 91% for urban ADSL connections. This is also the case for fibre packages, however, the difference between rural and urban performance was much smaller, with rural fibre connections receiving an average of 90% of their maximum speed at peak-times, compared to 91% for urban connections.  

Figure 1.6       Average actual urban and rural ADSL and FTTx download speeds

Source: SamKnows
Panel Base: Urban fibre 135, rural fibre 18. Urban ADSL 406, rural ADSL 169.

1.7 The extent to which download speeds are affected by peak-time network contention varies by technology

Download speeds typically vary by time of day and tend to fall during peak times, when a larger number of connections are being used and as a result of capacity constraints (contention) on the networks of internet service providers (ISPs) (Figure 1.7). For all the connection categories analysed, the peak time average actual speeds were lower than both the actual average maximum speed and the 24 hour actual average speed. In May 2014 the actual average speed across all connections was 18.0Mbit/s during the 8pm to 10pm weekday peak-time period, which was 89% of the 20.2Mbit/s average maximum speed and 96% of the 18.7Mbit/s 24-hour average.

ADSL1 and ADSL2+ connections capped at 10Mbit/s or less experienced the most variation between maximum  and peak-time download speeds, with average peak-time speeds (3.2Mbit/s) being 80% of the average maximum speed (4.0Mbit/s). ADSL2+ connections with a headline speed above 'up to' 10Mbit/s or without a headline speed performed better, with average peak-time speeds (8.1Mbit/s) being 91% of the 8.9Mbit/s average maximum speeds, while for cable broadband connections this proportion ranged from 83% for ‘up to’ 152Mbit/s connections to 90% for ‘up to’ 30Mbit/s connections (for ‘up to 60Mbit/s cable connections it was 89%).

FTTC connections were less affected by the peak-time contention than the cable connections of all speeds. The peak-time download speed on ‘up to’ 38Mbit/s FTTC connections was 33.0Mbit/s, 95% of the maximum average speed and 99% of the 24 hour average. For ‘up to’ 76Mbit/s FTTC connections the peak-time download speed was 57.8Mbit/s, 94% of the maximum average speed and 99% of the 24 hour average.

Figure 1.7       Variations in download speeds by the time of day: May 2014

Source: SamKnows measurement data for all panel members with a connection in May 2014.
Panel Base: 956
Notes: (1) Data have been weighted by ISP package and LLU/non-LLU connections, rural/urban, geographic market classification and distance from exchange to ensure that they are representative of UK residential broadband consumers as a whole; (2) As sufficient sample sizes were not available for consumers on packages of ‘up to’ 2Mbit/s or less, data collected for these packages in April 2009 has been factored in, in proportion to share of all connections in May 2014; (3) Data collected from multi-thread download speed tests.

1.8 The average actual upload speed of a UK residential fixed-line broadband connection was 2.4Mbit/s in May 2014

Broadband advertising tends to focus on download speeds. However upload speeds are important for a subset of the population, such as those sharing large files and using real-time two-way video communications. Our research shows that average actual upload speeds increased by 0.1Mbit/s (5%) to 2.4Mbit/s between the six months from November 2013 and May 2014.

Figure 1.8        Average upstream broadband speeds: May 2014

Source: SamKnows measurement data for all panel members with a connection in May 2014.
Panel Base: 956
Notes: (1) Data have been weighted by ISP package and LLU/non-LLU connections, rural/urban, geographic market classification and distance from exchange to ensure that they are representative of UK residential broadband consumers as a whole; (2) Data collected from multi-thread speed tests.

The full document is available below