A recent SEO study set out to discover if there was a negative impact on ranking on sites that were on shared hosting.
Google’s John Mueller offered his opinion of the test results.
Research on Web Hosting and Ranking
An SEO company set out to do a long term study on whether shared hosting negatively affects rankings.
The study authors reveal they created the study from concern of “potentially harmful SEO effects” from shared hosting.:
“Concerned about the potentially harmful SEO effects that hosting a website on one of these shared hosting options could have on the websites of thousands of business owners, we decided to run a technical SEO experiment to find out what effect, if any, it had.”
The idea they were testing is if low-cost hosting attracts low-quality websites that can negatively impact other sites hosted on the same server.
The report authors suggest that Google could use the presence of spammy sites on a shared server as a negative quality signal because they were all in the same neighborhood.
Here is their hypothesis:
“With cheaper shared hosting solutions usually attracting many lower-quality websites (like spam and PBN websites), we wondered if it is possible that Google’s algorithms sees using this kind of hosting as part of the pattern of a lower-quality website.”
So it appears that the authors are testing a hypothesis that is not based on any actual evidence or statement from Google.
It’s just an opinion expressed without basis and, as you will see, it’s also based on a misinterpretation of what a “bad neighborhood” actually is.
The article went viral on Twitter and Facebook, with many prominent SEOs promoting it.
Whoa – should be shared 1000x. Can cheap hosting w/ low-quality, spammy "bad neighborhoods" actually impact Google rankings?
"Long Term Shared Hosting Experiment" – Surprising new study+data from @rebootonline
Friends don't let friends buy shared hosting https://t.co/XAQHwgyG00 pic.twitter.com/2t6G3HfoyJ
— Cyrus (@CyrusShepard) September 18, 2020
Shared Host Study Methodology
The study authors went to great lengths to shield their websites from any outside interference.
They assured that nothing affected the results of the study other than the natural events of releasing them all at the same time for indexing by Google.
The methodology relied on an old research approach that uses a made-up keyword that has never previously existed online in print. The search term the created is the made-up artificial keyword, hegenestio.
Results of the Shared Host Research Study
The results of the study confirmed their pre-existing hypothesis. They discovered that shared hosting can negatively affect search rankings.
Study Authors Claim “Detrimental Effects” from Cheap Shared Hosting
According to the research authors:
“The results of this experiment suggests that cheap shared hosting options can in fact have a detrimental effect on the organic performance and rankings of the websites hosted there if your website ends up being hosted alongside lower-quality and potentially spammy ones (providing all websites being observed are otherwise on a level playing field).”
Study Authors Disavowed Their Own Findings?
Curiously, while the study authors first say that shared hosting “can in fact have a detrimental effect on the organic performance and rankings,” they actually disavow the bold statements in a following paragraph by stating:
“It is important to note that these results don’t show what effect the type of hosting you use when setting up a website would have in an actual SERP for a keyword with real competitors.”
What’s going on, right?
They first say that cheap shared hosting “in fact” can have a detrimental effect and then they say that the results “don’t show what effect” hosting will have in actual search results.
It kind of sounds like they’re trying to sound the alarm that “cheap hosting” can wreck your rankings while also saying, well maybe not.
Report Claims Benefits of Dedicated Server & IP Address
Then in a subsequent paragraph they seemingly contradict themselves again by asserting that the study does “show the effect the type of hosting you use” by writing:
“Hosting your website on a dedicated server and IP address has many benefits and now, according to the experiment data, ranking higher in the SERPs could very well be one of them.”
The SEO research authors hedge their statements with words like “could very well” which makes the statements less conclusive.
The statements that shared hosting “can in fact have a detrimental effect” on rankings and in a follow-up paragraph that the study “results don’t show what effect” shared hosting will have in “actual SERPs” seem contradictory.
Google’s John Mueller Comments on Study
Google Webmaster Trends Analyst John Mueller tweeted that the research was flawed and did not reflect how Google actually ranked websites.
Mueller asserted that research based on artificial websites (a reference to the made-up keyword used in the research) are flawed.
“Artificial websites like this are pretty much never indicative of any particular effect in normal Google Search. …it’s not useful data.
Host where it makes sense for you.”
Artificial websites like this are pretty much never indicative of any particular effect in normal Google Search. It's a cool experiment, a good write-up & analysis, and I love it when folks experiment like this, but it's not useful data.
Host where it makes sense for you.
— 🍌 John 🍌 (@JohnMu) September 18, 2020
Cyrus Shepard challenged Mueller to back up his statement:
“Could you present us with more useful data then to back up your claims?”
John Mueller tweeted this response:
“Are you saying that what works for made up keywords on artificial websites will work for a website in an active niche?
That seems like quite a stretch. (Even aside from the technical aspects of “what is shared hosting anyway” — AWS is shared hosting)”
Google’s Mueller followed up with this tweet:
“I’m not aware of any ranking algorithm that would take IPs like that into account.
Look at Blogger, there are great sites hosted there that do well (ignoring on-page limitations,etc), there are terrible sites hosted there. It’s all the same infrastructure, the same IP addresses.”
Problem With SEO Studies Based on Made-up Words
A problem with studies that use nonsense words is in the way Google’s algorithms responds to non-existent keyword phrases is different from it responds to actual keyword phrases.
Google’s algorithms attempt to understand the meaning of a word. It relies on language to rank webpages.
Without language or understanding, Google’s algorithm can’t kick in the way it normally does.
The only thing left is for Google to resort to simple word matching. And that’s not Google’s regular ranking algorithm.
So it’s kind of stretch to assume that the results of a test that uses a made-up word is going to reflect what Google’s regular ranking algorithm does.
That’s an important caveat against the usefulness of the study.
Related: SEO-Friendly Hosting: 5 Things to Look for in a Hosting Company
Research Authors Definition of Bad Neighborhood
A problem with the test is that it appears that the authors may misunderstand what the SEO term “bad neighborhood” means.
According to the article:
In the context of where a website is hosted, a “bad neighbourhood” is commonly referred to as a host, IP address and/or other virtual location where a collection of low-quality, penalised and/or potentially problematic websites (e.g. porn, gambling, pharma) are hosted.”
That is an incorrect statement.
A bad neighborhood has always been understood within the context of linking patterns and the spamminess of the sites being linked to.
In the past, Google’s Webmaster Guidelines referenced bad neighborhoods within the context of linking patterns:
“Don’t participate in link schemes designed to increase your site’s ranking or PageRank. In particular, avoid links to web spammers or “bad neighborhoods” on the web, as your own ranking may be affected adversely by those links.”
Former Googler Matt Cutts used the phrase “bad neighborhoods” himself and it was also within the context of linking patterns.
For example, in this blog post from 2006, he explains why some sites were dropped from Google’s index following an update called Big Daddy. Matt uses the phrase, “bad neighborhood” in the context of links.
“People have been asking for more details on “pages dropping from the index” so I thought I’d write down a brain dump of everything I knew about, to have it all in one place.
…The sites… were sites where our algorithms had very low trust in the inlinks or the outlinks of that site. Examples that might cause that include excessive reciprocal links, linking to spammy neighborhoods on the web, or link buying/selling.”
The phrase “bad neighborhood” is commonly understood to be within the context of linking patterns, not web hosting or IP addresses.
The context is never the web host or IP address, contrary to the misinformed assumption made by the study authors.
This means that the underlying assumption by the study authors, that web hosting is related to “bad neighborhoods” is incorrect.
Cyrus Shepard Changes His Mind About His Tweet
Cyrus Shepard changed his mind the next day about his original tweet. I admire people willing to change their minds.
On the following day he tweeted:
“So a bit of a mea culpa…
@JohnMu was 100% correct to say we shouldn’t conclude “all shared hosting is bad”
That’s not what the data says, and my tweet extrapolated too far (most shared hosting is fine – you’d still likely avoid the cheap/spammy operators)
The data though…”
Research Studies Can Be Problematic
There were many problems with the research, beginning with the fact that the authors appeared to have a bias in favor of a specific outcome.
Having a preconceived opinion can set up a situation called Confirmation Bias, which is a flaw that causes people to see evidence of what they believe and ignore evidence that doesn’t conform
I asked Jeff Ferguson (@CountXero), Adjunct Professor at UCLA and Partner & Head of Production at Amplitude Digital (@AmplitudeAgency) about this study and he offered this opinion:
“Hosting and the bad neighborhood theory has kicked around SEO circles for years but I never felt like it held much water.
While the Reboot test went to some extraordinary lengths to exclude known possibilities such as site speed, I still think it suffers from confirmation bias.”
I highly recommend an article that Jeff authored on the topic of SEO research, Do We Have the Math to Truly Decode Google’s Algorithms?
Insights and SEO Studies
SEO studies can be problematic. Although the SEOs behind this latest study went to great lengths to firewall the test to isolate it from anything that would bias the results, the basis of the study itself was flawed.
Firewalling the test didn’t matter because SEO studies based on non-existent keywords can not offer insights into how Google’s algorithms actually work, for the reasons I outlined above and as was confirmed by Google’s John Mueller.
My professional experience with shared hosting is that it generally has no effect on the ability of a site to rank. That’s not my opinion, that’s a statement of fact from my actual experience, over twenty years of experience.
The only time shared hosting becomes problematic is when a site becomes very successful. That’s why I also pay for dedicated servers.
Shared servers cannot handle ultra high traffic websites. I can say that with confidence because I’ve been kicked off of a shared server because my site’s high traffic slowed the server to the point that everyone’s site was throwing 500 errors. So you know… speaking from experience.
If you’re just starting out or don’t have excessive traffic, it’s perfectly fine to put it on a shared web host then upgrade when you outgrow it.
In my professional experience, based on what I know, I agree with what Mueller suggested, that publishers should host “where it makes sense” for them.
You can read the shared hosting SEO study here.