Internet Speed Test: Why ISPs Should Self-Host - BNDWDTH

Internet Speed Test: Why ISPs Should Self-Host

March 8, 2019

The need for speed: we've all felt it when it comes to our broadband. If there are questions about the connection, an internet speed test is sought out. More often than not when that need arises, users will fire up the Google machine. The first result? Ookla Speed Test. It's a great resource but not necessarily the most accurate one.

In order to serve customers better, an Internet Service Provider (ISP) should self-host their own speed test. We're here to tell you why. Read on!

How A Speed Test Works

Before laying out why a self-hosted speed test is a good idea, let's take a look out how a third party speed test works. A speed test like the one Ookla operates uses Transmission Control Protocol (TCP) testing with a HyperText Transfer Protocol (HTTP) fallback for maximum compatibility. TCP defines how to establish and maintain a conversation between nodes so data can be exchanged over a network. Both of these protocols check latency, download and upload components.

Many times a "ping" test result will also be displayed which primarily measures latency. This estimate is calculated by sending HTTP requests to the testing server and measuring how long it takes to get a response. After this approximation, an actual speed test is run. The test saturates the line to get a more accurate measurement of how much data can be downloaded in 10 seconds. The goal is to retrieve this information without taking too long.

Inconsistent Results

Different factors come into play that affect the accuracy of a test. There are six areas to take into account: sample files used, server location, your activity, time frame, your gear and test popularity. 

  1. SAMPLE FILES - A third party speed test uses sample files that are vastly different from how someone may actually browse the internet. Gaming online or streaming HD video uses much more bandwidth (One of our favorite words!) than these simple files.
  2. SERVER LOCATION - At the very core, the test is simply sending files from the user's computer to a server in a different location. The further that server is from the computer running the test, the longer it will take for the reading. You may start to see why a test being served via the local ISP can be more reliable.
  3. YOUR ACTIVITY - Is someone in the house streaming Michael Jackson's Thriller while the test is run? Most households have several devices connected to the WiFi network and they are all pulling from the same available bandwidth. The more people on at the time of the test, the slower the result will be.
  4. TIME FRAME - Not only will the usage in the home affect it, but also those that share the same internet provider can also slow down speeds. Certain times of the day will be more congested and result in a different reading than say, before breakfast.
  5. YOUR GEAR - If a user has an older modem or rickety computer, the speed test results will show it. A hardwired connection may be more reliable than WiFi depending on the device location. In some cases a hardware upgrade may be the solution for faster speeds more so than a bigger internet package.
  6. TEST POPULARITY - Similar to home network activity and shared ISP activity, the number of people using the test at one time can affect results. In 2018, 115 million U.S. residents tested their connection speeds on Ookla during a six month span. That's over 630,000 tests a day. It might just create some bottlenecks.

The Case for Self-Hosted Speed Tests

If you've been paying attention, you may already be able to make the case yourself. We'll reinforce the learning as it's good practice any way! No matter where the test is hosted there will always be variables that will affect the outcome. A self-hosted speed test eliminates some of the factors which will serve the ISP and the customer better.

It's Your Network.

Having the test on your server, living on your network means that it is a more accurate representation of how fast (or slow, yikes!) the internet is. This is the single most important factor when a customer needs to check the connection speed. The challenge is to make sure they are using your speed test instead of just googling a third party application. Educating your customer base about the difference is something your marketing team will need to do. Of course, if you need assistance with marketing, we know somebody ... it's us. Surprise!

Proximity Wins!

On a related note, proximity is another important part of getting better speed test results. Previously, we discussed how a third party provider will try to locate a server that is close to the tester. It's not a foolproof system as the internet, like water, is going to take the path of least resistance. When someone runs a test in your coverage area there's a much greater chance that your test will have a direct path. The "other guys'" test may be rerouted through California to get back to Iowa or wherever you happen to call home.

It's Less Crowded.

As an independent or regional ISP,  you are probably not serving 100 million unique requests in 6 months. Of course, this means that the pipeline is less crowded and your customers have a clearer, more open path when running their speed test. Voila, a more accurate result!

One other factor to keep in mind: someone outside your network or on a smartphone that visits your self-hosted solution will be measuring the speed of their current provider. There are a lot of factors, people!

Despite the many nuances of trying to pin down a good result on a speed test, a self-hosted option helps the ISP give their customers a more accurate reading. If you need help getting it set up, we can do that.

BNDWDTH is proud to develop marketing tools and strategies that are specifically for broadband providers, cable TV operators and telephone companies. We're built to help you! Contact us with any questions or comment below.


About the Author


BNDWDTH is a marketing agency built for communications service providers. We help cable TV, broadband and telecom providers increase market capacity.


Leave a Comment:

All fields with “*” are required

Leave a Comment:

All fields with “*” are required