Faster Web Servers Mean Higher Ranking Claim Options

[xyZed]

I've just read the claim from a web hosting company that having a faster server increases your search engine rankings. Frankly it seems a ludicrous claim but this is one of the top uk web hosting companies - what do you think?

"This month we've completed a study into the ties between search engine optimisation and server speed.

Our research culminates in a whitepaper - which is available for download from the UKFast website.

The faster your web server is, the higher in the search engines your site will appear.

Search engines such as Google reward businesses that take their sites more seriously by boosting their position in the rankings."

[Respree]

I think its a marketing strategy that takes advantage of a claim that can be neither proven nor disproven.

I wonder what they will do when customers start demanded their money back when their rankings having magically not improved.

Caveat Emptor.

This post has been edited by Respree: Mar 31 2008, 04:34 PM

[cre8pc]

So. If you have a REALLY bad web site, but your server is fast, Google will rank your site higher?

I find that impossible to believe and I bet Google does too

[EGOL]

Two cents worth of speculation....

If your site is on a very slow server you will have more short visits (people leave because they are tired of waiting). If search engines consider the number of short visits in their rankings algo then your rankings could be reduced. Also each short visit are ads that are not clicked, products that are not purchased, page impressions not made, and links not earned.

So, if you have a high traffic site the cost of better hosting could be very small compared to the loss of revenue. It is a big mistake in my opinion to skimp on costs by tolerating crappy hosting.

"Search engines such as Google reward businesses that take their sites more seriously by boosting their position in the rankings."

I don't know if you can tie this directly to server speed, however the concept is 100% valid.

Visitors judge Google on the basis of their performance - by providing a SERP full of great websites. Therefore it only makes sense that they will pass that on to the content that they include on their SERPs.

One more comment, then I'll shut up.... I moved my highest traffic site to a dedicated server and got a fast one. I did not know what time of day that they were going to switch the site, however, I knew it had happened when the speed of a pageload knocked me out of my chair.

This post has been edited by EGOL: Mar 31 2008, 07:53 PM

[projectphp]

Even simnpler answer:
Better sites can afford better, faster hosting. Self fulfilling prophesy!

[coolguy27]

I think fast server can contribute to high rankings....


Since we know that Google have standards in speed of downloads of the site...

This post has been edited by coolguy27: Apr 15 2008, 05:38 AM

[Autocrat]

The claim itself is 'true' ... it's how they say it.

It's the whole marketing spin.
Instead, what it should say is something alike
"Having very slow server response times and lower output bandwidth can results in Search Engines giving up attempting to crawl and index your site, causing lower results"

Instead, they have taken that, turned it round and added some sparkle... marketting

[Joshua Sciarrino]

They are such good marketers that we're discussing this and thinking about them.

Heck, I had a dream about them last night, 'Without this company I'll never get first page on google'. Then I woke up and...just remembered it was just a marketing sceam...i mean dream....

This post has been edited by Flying Monkeys: Apr 15 2008, 09:31 AM

[Autocrat]

Yup.. .as much as I tend to dislike Marketing (due to the nature of some of the folks involved, and the general deception)... it is rather impressive and pervasive, no matter how you look at it (sometimes I do find myself in awe... either from sheer audacity to grudging admiration).

Still, on topic... it can make a difference, but more from the negative/slower end rather than the positive/faster end of things.

[Joshua Sciarrino]

Marketing has a way to either draw you in or push you in.

I think the opposite is true though, about faster server, better search rankings. Why? The line of logic and it statistically makes since.

A website thats down 50% of the time will NOT rank as high as a website with 99.9% uptime guarantee. Follow the line of logic EGOL gave. I think it's solid.

Their are always outliers but generally speaking I think this is a standard.

Why create a 24/7 business if the door is closed half the time? It's aggravating for the prospect clients. No one likes people that aren't professional and a professional website is expected to be up all the time, let alone optimized, and fancy graphics.

[Ron Carnell]

About ten years ago, I was driving a 26-foot Uhaul truck, with my car attached behind it, from California to Michigan. It was a brutal trip, especially the first several days when I was still in the mountains. There were times, many and frequent, when I couldn't get over thirty miles per hour. Too much weight and too little power. I was constantly worried the police were going to pull me over for going too slow. On most American highways the minimum allowed speed is 45 mph.

That's sort of what we're talking about here, I think. The minimum speed allowed. If your server is too slow to serve a page, the visitors are going to go away -- be they human or arachnid. Get above that speed and the issues change. Unlike your human visitors, the spiders are facing a quota. They want to get X number of pages from your site today. If your server is too slow to give them X number of pages, you're going to have fewer pages indexed than you deserve. Google's Webmaster Tools will generally tell you if this is happening. Go to Tools / Set crawl rate and scroll past the graphs. If the "Faster" option is disabled, it means your server is already as fast as Google needs it to be. You could upgrade to a Maserati web server and Google still wouldn't index any more pages. The link pop of your site is the funnel, not server speed.

Beyond indexing, I don't believe the speed of your server plays any role at all in rankings. It can't, not if search engines are truly concerned with relevancy.

Google can't tell how fast your server is. It can only tell how fast a page is delivered when a request is made. A slow response might mean you're still running on a 486, but it could just as easily mean you have the most popular web site on the Internet. There's no technical way to tell the difference (save, perhaps through inference), and the engines sure aren't going to penalize anyone for being popular. Even from the visitor's viewpoint, speed is always going to be secondary to relevance. It might take me ten minutes to bring up my bank account, but I'm going to wait however long it takes because the relevance of that page is extraordinarily high for me. Good search engines are going to concentrate on relevance, not on speed, not on garish graphics, not on poor usability. Those are all reflected in link popularity any way.

If your page can be indexed, is highly relevant to the search, and has the link juice, it's going to be ranked well regardless of server speed.

[SEOigloo]

Because things like time spent on site are a very likely part of the algorithm, quick loading pages might be seen as a ranking factor, by association, simply because a slow loading site is likely to frustrate people who will not want to spend as much time on the site.

The marketing of this is clever, but it is basically true, for several reasons like the above.

Good question!
Miriam

[Ron Carnell]

Because things like time spent on site are a very likely part of the algorithm ...

How does Google know how much time I spend on Cre8, Miriam? Or on any page I find in the SERPs?

[Autocrat]

...Ron Carnell...
It appears that if the bots cannot access your site on several attempts, then yes, it does affect your rankings.
I've known sites to drop due to it... solely because they could not index.
(think about it... if you cannot read the book, how can you decide how good it is?)

I doubt if it is immediate... but it does seem to happen.
Additionally, the more often it happens, the longer the time inbetween the 'looks' may also increase as well (it becomes less and less interesting).



As to telling the time.
Well... immediately to mind is Analyitics etc.
Secondly, maybe the google serps results pages use JS too (for detecting multiple clicks?).

There are probably other ways as well... but those were the first two off my head.

[Ron Carnell]

It appears that if the bots cannot access your site on several attempts, then yes, it does affect your rankings.

Not in my experience, Autocrat. In my experience, it affects your indexing, much as I said in my earlier post. It may seem like a trivial distinction, but I think it's an important one. If your page is in the index, you should be confident its rank won't depend on server speed or any other issue not related to search relevance.

As to telling the time.
Well... immediately to mind is Analyitics etc.
Secondly, maybe the google serps results pages use JS too (for detecting multiple clicks?).

I don't use GA on any of my sites. I'd guess about 98 percent of the Internet also doesn't use GA. And, while Google does indeed track SERP clicks with Javascript, that in no way tells it how long I stay at any of the sites that I routinely open in a new browser tab. You didn't mention the Google Toolbar, but again, I don't have it installed and I'm not all that atypical from billions of other surfers.

I will certainly grant that Google can track some surfers and some web sites. But these are mathematicians we're talking about, and I personally don't think the numbers are statistically significant enough to warrant overt inclusion in the algorithm. The absolute last thing any search engine would want to do is rank sites using GA over sites not using GA at the risk of relevance. That would be, well . . . evil?

[Autocrat]

Well I'll agree disagree on the server speed affecting things

It would be interesting to know jsut what G is tracking and how (possibly adsense stuff as well... maybe?).
As t oevil... don't make me laugh - considering some of the results in the SERPs and how they got there... I can think of much worse sins/crimes than ranking sites based on somewhat questionable statistics

[projectphp]

What part do you disagree with?

My experience, is like Ron's: slower == less pages, but those pages indexed do not rank worse.

The other part about time on site is intent. How long do i need to be on this page for?
http://www.google.com/search?q=time+in+buenos+aires

Would 15 seconds make the page better?

There are bett signals, for an SE algorithm, than time on site, which is a better metric for marketers to messure success.

[Autocrat]

Well, I'm disagreeing with the idea that slwoer servers, causing timeouts and bots being unable to crawl the entire site, results in a potentially lower ranking.
I'm basing that disagreement on experience.
A poorly setup server hosting sites, one of which is not ranking anywhere near what it should've been (or atleast, as hoped for considering hte efforts put into it).
Fix the server.
Increase the response times.
No more 404's or time out errors or reports of being unable to access the page/site.

Site goes up in the rankings.


No changing of server (so nothing to do with the IP block/range // penalties).
No changing of Domain name (so nothing to do with links // penalities).
No changing of on-site/on-page optimisation/content (so benefit/loss due to that either).
Checked yahoo... no change of links (so nothing altered from that factor).
No alteration of complimentary files (robots.txt, .htaccess or xmlsitemap).

The only thing to be changed was the server setup, increasing the response times and the bots stopped getting errors.


Further, I'm not the only one who perceives that Google may rank a site based on it's entirety... which kind of requires that the site gets indexed... not just a tiny part of it.


So, yes, I am assuming that it is server response times and the lack of errors that caused the increase.
I could be wrong... there may be some factors that I've missed.
Could make for an intresting study.

[Ruud]

Well, I'm disagreeing with the idea that slwoer servers, causing timeouts and bots being unable to crawl the entire site, results in a potentially lower ranking.

As you then go on to explain why, according to you, slow servers do lower rankings I'm left wondering if this is a typo or that you're disagreeing just to disagree...

But let's assume you really did mean to make the claim that server speed is a variable used in the relevance calculation that is the ranking of search results.

• Page A has a 100% relevancy score for query Z. Page B coming from that same site, same server, loads painfully slow. You suggest this would lower the relevancy of page A by which factor?
• Page A has a 100% relevancy score for query Z. Page B from a different site but on the same server never ever loads -- this lowers the ranking of page A?!
• Page A by my competition ranks above me. I rent space on the same server and artificially delay the server response. This lowers the ranking of my competitors web pages by....?
• If a loss in server response time equals loss in ranking (for that page, site or server ... or however you would construct this idea) than what is the base? The fastest server on the web? If not (because that would be insane) you would use a median and... boost the ranking for pages which are less relevant but served faster? Seriously?

Further, I'm not the only one who perceives that Google may rank a site based on it's entirety... which kind of requires that the site gets indexed... not just a tiny part of it.

But you do realize that at no time Google will be aware what "entire site" constitutes?

I'm basing that disagreement on experience.

Yet you share an anecdote, not an experience, let alone a semi-serious test. Can you repeat the results? Can you take a page and lower and higher its relevance by adjusting the speed with which it is served? Time and time again, on different sites, different servers? If so, I'm looking forward to your blog post or white paper on it and can guarantee you free publicity, starting with an interview on Search Engine People!

there may be some factors that I've missed.

Yes, given that you might test this, I thought I would indeed touch upon that.

No more 404's or time out errors or reports of being unable to access the page/site.

To clarify: we're talking about server response time -- not about absence of server response. That also means that 404's, a status code issued by a web server in response to a request it temporarily cannot answer, shouldn't be included. For one, they are a server response. Second, adding a bunch of errors into the mix would muddy your tests, right?

On the other hand, makes for an interesting twist on your claim; what if a server is really fast in serving up 404 status codes? would that boost the relevance of the site hosted on the server? Hmmm....

The only thing to be changed was the server setup, increasing the response times and the bots stopped getting errors.

Dude.... how can you say this in one sentence and not scream at your own contradiction? The "only" thing that changes is the server response time ... oh, and there are no more errors. Duh....

<shakes head> This information is so way off it should have an SEO Health Warning on it.

This post has been edited by Ruud: Apr 16 2008, 06:55 PM

[Joshua Sciarrino]

Great discussion. Be careful of any attacks on one another though. I've made up my mind on the issue but am still loving this discussion and I don't really have any in-depth notes to share about the topic matter but I do share a concern, let's not get too heated about this.