Home Tech Internet Speed Testing May not look What it Appears

Internet Speed Testing May not look What it Appears


A number of online internet speed testing services may claim to produce completely authentic speed tests ensuring 100 per cent accuracy but the reality appears to be far from reality.

Presently, several internet speed testing apps are operating around the globe with hundreds of millions of internet users worldwide extensively relying on these sites to test the speed of their internet connections.

Contrary to the widely believed perception that the credibility of relevant speed test results are beyond doubt, the reality is different than meets the eye.

In Pakistan, Ookla is one of the online internet speed testing services, popular among the common internet users. However, like other internet services, the speed testing results produced by Ookla are also deficient and valid questions can be raised over their authenticity keeping in view a number of technical factors that are mostly ignored by a common internet user. These are detailed below:

Speed Test, a subsidiary of Ookla, functions to test linkage between an internet operator and their customer, not its speed to load website or download a file. If an internet user actually wants to test his/her internet speed then they should always try a server located remotely, in countries like Netherlands, Singapore or United States. Needless to mention that these are the countries where most of the internet data travel to and from.

Secondly, it is a common practice on part of internet operators to host a speed test server within their own premises. This allows the operators to ensure that the internet users unknowingly select default server when testing their internet connection speeds that shows the speed promised by the operators.

Thirdly, international bandwidth is costly and operators want to make money while serving ‘lightening speed’ to internet users. They buy 10 GBPS pipe and sell 100 MBPS connection to 10000 people. They also install speed test server at their own datacentre so that internet users do not complain of low speed. Hence, when the internet subscribers actually try to download stuff from remote location, they get poor speed. This is mainly the case with Speed Test that claims to show ‘real’ internet speed results.

Ordinary internet users often mistake internet speed with bandwidth. Internet service providers claim that their connections are as fast as 35 MBPS or that their speeds are 20 per cent faster than their competitors. In reality, however, a 25 MBPS connection has little to do with speed and more to do with how much data an internet user can receive every second.

True internet speed is a combination of bandwidth and latency. Simply put, latency = delay. It is the amount of delay (or time) it takes to send information from one point to the next. Latency is usually measured in milliseconds or MS. It is also referred to (during speed tests) as a ping rate.

Technically put, bandwidth has to do with how wide or narrow a pipe is and latency has to do with the contents of the pipe; how fast it moves from one end to the next.

Lower the latency the faster the internet speed and latency is also affected by the distance between subscriber and mobile tower. More distance from the tower will result in higher latency resulting in low internet speed besides having high bandwidth. It is similar as one operator having 100 towers for 1 million customers and the other one having 50 for same.

Speed Test has no score for latency that is very important part of internet speed test. In realistic terms, low latency with low bandwidth is better than higher latency with higher bandwidth.

Greater the number of customers lesser the bandwidth. When telecom operators launched 4G services, the respective speeds they offered were higher. Gradually, the speed got to get slow with the passage of time.

In Pakistan, a sample size of test conducted for Q1-Q2 is very small in comparison with the number of 3G and 4G users.

Given that the number of 3G and 4G subscribers in Pakistan is 59 million, only 739,728 used Ookla’s Speed Test in over six months. Interestingly, only 164,475 ‘Unique Users’ are part of this test. Unique Users means the users who are not repeated.

Speed Test score calculated that 90 per cent of the final speed score is attributed to download speed and the remaining 10 per cent to upload speed. This implies that download speed gets more score than upload speed that is another flaw in this test.

More importantly, as per Ookla’s own laid down professional criteria, it does not conduct speed tests of samples below 3 per cent. Of 59 million 3G and 4G users in Pakistan, Ookla conducted internet speed tests of 739,728 users which is only 1.25 per cent, less than half of Ookla’s fixed criteria of 3 per cent. Glaringly, this is against Ookla’s own methodology.