Do Hosts Slow Down Or Put You On A Slower Server After Some Time? (3.5 mln speed test, 5 years,15 hosts)

Share the knowledge...Share on Facebook
Facebook
Tweet about this on Twitter
Twitter
Share on LinkedIn
Linkedin

hosts slow down - title

You’ve probably heard of or even experienced yourself that once you start using a hosting it performs pretty well at the beginning. But after several months it looks like you are moved to a slower server.

I decided to use years of my non-stop speed monitoring of 15 hosts to analyze this hypothesis.

 

What I’ve done to analyze the speed of the hosts

  1. Took monthly average speed data for the 15 hosts I had been monitoring from February 2016 to November 2020 using the monitoring tool monitis.com. This makes almost 5 years of data. I started to monitor each hosting in a non-stop manner right after I bought the hosting plan (the monitoring methodology is here). Totally there was about 3.5 mln speed tests done.
  2. Built speed charts for each hosting using average monthly data (see table data here).
  3. Visually analyzed how speed changes to see if it is true that hosts put you on a slower server after several months as you buy a hosting.

 

Something to say before going into the analysis

  • The hosts that I’ve been monitoring are considered to be good and great hosts. They are not the hosts known for unethical behavior. Making a conclusion for any hosting in the world analyzing only my hosts is not correct. In other words, the set of hosts that I am analyzing is not representative.
  • Having said the above, why do I still making this research? The point is that I assume you make some research before choosing a host. And if you filter out unethical hosts, then my set of hosts will become more representative. This way it makes sense to analyze the speed of my hosts then.

 

What I’ve seen and my thoughts

Here’s a chart with monthly average hosting speed values displayed (each chart represents about 5 years of data, clickable):

hosts speed charts 58 months

  1. To a certain extent it indeed looked like many hosts performed well at the beginning and then the performance degraded. At least 11 host out of 15 behaved like this.
  2. Assuming that most hosts (11 hosts) behave in a very similar unethical way looked like a conspiracy theory. Although it still could be true (anything can happen in our life), I wanted to look for some other explanations.
  3. Here’s one of the reasons why this hypothesis (i.e. these hosts move your account to a slower server after a while) does not look like a true one. The point is that the clear pattern of slowing down takes place only at the beginning of the monitoring whereas during these 5 years I bought hosts again as a new user. But there were no such similarly significant speed changes any more like the one at the beginning.
  4. This made me think that the speed increasing anomaly at the very beginning was not totally connected with hosting speed. Perhaps it was due to issues on a monitoring tool’s side. Let’s try to check this.
  5. I picked out the peaks (clearly noticeable slow months) soon after the beginning of the monitoring. And tried to pick out just one month. It appeared like it was the 9th month starting from the beginning. And I put it on the chart (see the orange vertical lines on the charts).
  6. So, I assumed that during about the first 9th month from the time when I started using the monitoring service the monitoring service itself showed slower results within time. Or in other words, at the beginning it showed better speed results than it should have to.
  7. One of the arguments why monitoring service may change its algorithms to measure websites speed is that there are multiple speed testing tools which show different speed for the same website at the same time (see my research here on this topic). Moreover, different speed testing tools show like the same website performs differently within time. In other words, speed testing tools are not super precise when testing website speed. But at least they give data for analysis which is really much better than a simple guess work or sporadic single speed tests (here’s my research why single speed tests are very unreliable).
  8. At the same time not all hosts clearly show the pattern of slowing down from the very beginning and having the slowest performance on the 9th month (which could be a very solid argument of the monitoring tools being in charge of this anomaly).
  9. The above point drives home another assumption that some host (which did not show the patterns of slowing down at the beginning) in fact increased their speed. The monitoring tools did not reveal that because it was assumingly slowing down the results. But is it 100% true? I can’t say so.
  10. The final argument why the host (at least the ones that I had been monitoring) likely do not put you on a slower server after some months is that the speed of many hosts fluctuate or even improve within time.
  11. The above point helps to assume the following. Some hosts may simply continue putting clients on a server and then when it appears that the server is too full (the websites on the accounts become too slow), the hosting moves some accounts to a new server or removing server abusers from a server (sharp improvement of speed). And a gradual speed improvement of website speed on a hosting may be connected with better server management.

 

Conclusion

This is a kind of a research when I expected to get other results when I just thought of starting doing this research. But the results I got made me look deeper.

I thought (and still think) that decent hosts do not behave in an unethical way like for example putting you on a fast server at the beginning and after sometime putting you on a slower server. This is simply not the way well-established and high-reputable hosts do. This is too mean and can be done only by hosts which play really risky (not the host in my list).

My monitoring results indeed showed good speed at the beginning and the speed was becoming worse gradually for about 9 next months in general. This could be a proof of unethical behavior of the hosts. But looking deeper reveals other factors why this pattern might take place.

One of the factors (at least in my case) is that there were probably some particularities with the monitoring system which changed its algorithms to measure website speed.

Another argument why a host might be not as mean as sometimes people may think is that hosts may not only show decreasing speed, but also within time the hosting speed increases (like in my examples, see the chart)! Yes, it also may be connected with particularities of a monitoring system. But there is no 100% reliable tool to know that.

After all, if you experience that your website became slower that it used to be, then it’s probably true. Whether it was because your hosting oversold the server (put too many users on the same server), or whether it’s your website that got too much under the hood and became slower indeed, you have to do something.

The first practical advice would be to contact your hosting support first and describe your concerns. Sometimes it may surprisingly help (the tech support look into the server and probably find the issues). It happens sometimes.

Another practical advice is to optimize your website (and this is a typical reply from a hosting actually to your speed concerns). Easier said than done though. Probably using a free plugin like this can help fix your speed issues to a certain extent you if you are not technical person.

Also, if it’s your website which has become too heavy, then the option is to go with a more powerful hosting plan (e.g. see this case suggesting affordable and very powerful hosting options).

You can download a PDF version of this article (536 KB):


Subscribe to my Free Researches
Work on your blog and small business more efficiently!

subscribe
BTW, I respect your privacy, and of course I don't send spam, affiliate offers or trade your emails. What I send is information that I consider useful.

Share the knowledge...Share on Facebook
Facebook
Tweet about this on Twitter
Twitter
Share on LinkedIn
Linkedin

Comments

  1. Wow Michael,
    This is a quite amazing result, looking at the graphs it would “appear” (at a quick glance) that your theory of misappropriating clients on slow servers after sign-up might be so…. BUT contrary to the information in your write-up it is not!

    I had a look at the company you used – seems like they are a well established entity on the web.

    Contrary to my first thoughts on your assessment, I can now see that the information is of high importance when it comes to exposing the good AND bad of hosts…. WHICH hopefully as your results become more public will subliminally send a clear message to the unscrupulous hosts to take notice!

    well done
    sincerely
    Bruce

    • Thank you for your input Bruce,
      Indeed all the inspected hosts are not bad ones at least. That’s why probably there are no patterns of decaying (and not recovering) performance.
      Although I assume that the performance on any host can decrease, but good and great hosts either do not let it happen for a long period of time and take appropriate actions to make the speed good again. Bad hosts do not take much attention to it for a longtime, keeping their servers oversold and overloaded.
      Among the host that I monitored, HostWinds probably shows this behavior for several months (the peak is at the 36th month) – the speed dramatically decreased, but then it recovered. I assume that there are hosts out there that do not recover the speed on their servers for a long time. At the same time, among my hosts there is another example – Glowhost, it showed constantly not great performance along all these years upto 202 inlcuding (it recovered only in 2021 according to my speed monitoring tool).
      So, there are multiple nuances to take into account 🙂

It's important for me to know what you think

*