European consumers are not getting the broadband download speeds they pay for. On average, they receive only 74% of the advertised headline speed they have paid for, according to a new European Commission study on fixed broadband performance.

Commission Vice President Neelie Kroes says, “This is the first time the difference between advertised and actual broadband speeds is confirmed by comparable and reliable data from all EU Member States.” There are significant differences in the European national markets, most likely due to advertising practices. Kroes says “Consumers need more of this sort of data to help make informed choices, so we will repeat the exercise. And we take these first results as further proof of the need for a real connected single market.”

Key findings in the study include:

Cable has the most reliable download speeds: The European average of 74% hides significant variation in the performance of different technologies. xDSL based services achieved only 63.3% of the advertised headline download speed, compared to 91.4% for cable and 84.4% for FTTx.

In absolute terms, the average download speed across all countries and all technologies was 19.47 Mbps during peak hours. FTTx services achieved the fastest speeds at 41.02Mbps. Cable services achieved 33.10Mbps, whilst xDSL services lagged far behind at 7.2Mbps on average.

The upload speeds are closer to their advertised speeds. Across Europe, the average upload speed was 6.20 Mbps, representing 88% of advertised upload speeds. FTTx services achieved the highest speeds by far, at 19.8Mbps. This is because many FTTx services provide an upload speed far closer to the download speed. Cable and xDSL services achieved a modest 3.68Mbps and 0.69Mbps respectively.

In Malta the results of the tests were as follows:

Actual download speed of xDSL technology as a percentage of advertised speed during peak periods. 86.0%, the average is 60.3%.

Actual download speed of cable technology as a percentage of advertised speed during peak periods. 95.4%, the average is 91.9%.

Actual download speed of xDSL technology during peak periods, 7,93Mbps, the average is 7.23Mbps.

Actual download speed of cable technology during peak periods, 21.56Mbps, the average is 35.1Mbps.

Actual upload speed of xDSL technology as a percentage of advertised speed during peak periods, 102.1%, the average is 84.6%.

Actual upload speed of cable technology as a percentage of advertised speed during peak periods. 98.6%, average 98.5%.

Actual upload speed of xDSL technology during peak periods. 0.59Mbps, the average is 0.71Mbps.

Actual upload speed of cable technology during peak periods 1.32 Mbps, the average is 3.34Mbps.

Latency is an important metric that is often overlooked by consumers. Latency (or more precisely, round-trip latency) is the measure how long a single packet of data takes to go from point A to point B and back again. In this study, round-trip latency is measured between the panellists’ homes and the nearest measurement server.

Given that every communication with Internet services involves the transmission and receiving of packets of data, latency affects everything we do on the Internet. It is particularly important for time-sensitive applications such as online gaming, video streaming and voice communications. The lower the latency, the more responsive the connection will be.

In Malta the results were as follows:

Latency of xDSL technology during peak periods 61.39ms, the average is 41.0ms.

Latency of cable technology during peak periods 46.20ms, the average is 21.7ms.

Packet Loss

Packet loss, like latency, is often overlooked when consumers are considering broadband products. This metric describes how many packets are lost as the traverse the path between the home and the computer they are communicating with. When packets are lost, the two parties involved in the communication will typically retransmit data in order to account for the loss. This takes time and when it reaches a certain level it will become very noticeable to users.

Realtime applications such as online gaming, video streaming and voice will again suffer the most from high packet loss.

Italy, Malta and Portugal stand out as the noticeable exceptions for delivering high levels of loss. Italy in particular is surprising, as multiple measurement servers were located in major cities in Italy, so the path between the user’s home and the server should not have been that far.

Packet loss of xDSL technology during peak periods, Malta was the highest at 1.89%, the average is 0.62%, Italy was 1.76% and Portugal 1.36%.

Packet loss of cable technology during peak periods, was 0.23%, the average was o.31%.

DNS Resolution Time and Failure Rate

DNS is a critical Internet service that allows you to turn hostnames (such as www.youtube.com) into IP addresses that your computer can communicate with DNS services are typically provided by the ISP in order to provide a fast, nearby service for their users to use.

However, a poorly performing DNS service can lead users to perceive noticeable delays. This is particularly noticeable when browsing the web, which is heavily reliant upon DNS.

In theory, a good DNS deployment should provide DNS resolution time and failure rates better than or equal to the latency and packet loss figures. This is because the DNS servers are typically hosted inside the ISP’s networks and therefore this traffic does not need to leave the ISP’s network.

In Malta the results were as follows:

DNS resolution time of xDSL technology during peak periods, 24.44ms. the average is 42.6ms.

DNS resolution time of cable technology during peak periods was 16.93ms, the average is 24.3ms.

DNS resolution failure rate of xDSL technology during peak periods 0.35%, the average is 0.40%.

DNS resolution failure rate of cable technology during peak periods is 0.14%, the average is 0.25%.

The web page loading times of xDSL technology during peak periods was 1.44s, the average was 1.69s.

The web page loading times of cable technology during peak perids was 1.28s, the average was 0.79s.

Results are based on peak time performance, which is defined as weekdays 7:00pm to 11:00pm (inclusive). These are the overall results of the study sample and do not refer to the actual composition of the broadband market across each country.

This study will run until end 2014 and two more annual measurements are planned. European consumers can measure their own ISPs performance by joining a community of volunteers across all EU27 countries plus Croatia, Iceland and Norway. Selected consumers will be sent a small device to plug into their home internet connection. This device will run a series of automated tests when the line is not in use. It will establish the speed and performance of their broadband connection. Go here for further information.

Speeds of broadband products are advertised as “up to XX Mbit/s”. This speed is called “advertised,” “nominal” or “headline” speed and it is what we see in the adverts. But there may be significant differences between the advertised speed and the actual speed consumers receive. Variations in download performance between Member States are primarily driven by the technologies that have historically been deployed in those countries.

Research shows that between 27% and 41% of European Internet subscribers claim that their download speed does not match that in the terms of their contracts. And almost half of EU subscribers claim that they sometimes experience difficulties accessing online content and applications due to insufficient speed or capacity.

This is the first study on broadband performance that covers all EU Member States, including, Croatia which will join the EU very soon, as well as Norway and Iceland, using the same approach. The project is being run by broadband performance testing specialist, SamKnows, which has already conducted similar projects in the UK and the US.

It is based on a methodology that uses hardware devices and provides the most accurate and independent results of internet performance regardless of access technology and home installation. This methodology has also been used by national regulators in the US, the UK, Brazil and Singapore.

The Phase 1 report was completed independently of Internet Service Providers (ISPs), however ISPs are invited to participate in Phase 2 and should contact ec@samknows.com for more information.

Results were taken in March 2012 from a panel of 9,104 participants. A total of 3,065,341,850 measurements were taken across 75,978,173 unique tests.

Permalink: Consumers are not getting the internet speeds they pay for