Are Subdomains better or worse than separate domains? Options

[bwelford]

Well we all watched bemused by the effrontery of David as he threw a few stones at Googliath and they all seemed to find their target.

What I'm talking about is what was discussed in the following two threads:

"t1ps2see" spamming Google big time!
http://www.cre8asiteforums.com/forums/inde...showtopic=37873

3 Week Old Site gets 4 Billion Pages Indexed by Google, Neat Black Hat Exploit
http://www.cre8asiteforums.com/forums/inde...showtopic=37878

So it was just a bit of fun and we all go back to work.

However is there anything to be learned from that incident. My current view is that subdomains are independent entities with Google just as separate domains are. In search engine optimization terms the two are equivalent.

Is that how everyone feels? Are there advantages or disadvantages of subdomains versus separate domains? Is it the same story with Google, Yahoo! and MSN? What do we really know on this?

[rustybrick]

It is that sub sub domains are the issue. I wrote about it at SEW http://blog.searchenginewatch.com/blog/060620-081946

[bwelford]

There's some interesting reading there, Barry. Thanks for pointing out that it's really sub-subdomains we should be talking about.

Let me rephrase the question with a simple example (I am using fictitious domain names here):
Suppose I have a vacation property rental website with the Home Page at www.rentals7235.com. I want to put up a web page for a rental property at Mont Tremblant in Quebec. Which of the two following URL's would likely work better?

• vacation.monttremblant.rentals7235.com
• www.monttremblantrentals7235.com

What are the positives and negatives of each? I guess the sandbox may come in here somewhere. Is this the same for Google, Yahoo! and MSN? How may things change in the future?

[Ron Carnell]

... a search on queer forum returns CraigsList 97 times out of the top 100 results.

Technically, while that's true, Barry, it's not much different than saying the search returns .org 97 times out of the top 100 results. We know, of course, that all .org domains are not treated as the same site. In precisely the same way, not all craigslist.org domains are treated as the same site either. Nor, in my opinion, should they be.

You can get similar results, if somewhat less dramatic, by searching for "Be a Guide" and looking for all the about.com subdomains (sub-subdomains are irrelevant).

Whether the issue is subdomains or sub-subdomains isn't relevant, in my opinion. The real issue is whether either should be considered independent entities. Should barry.tripod.com be lumped in the same barrel with ron.tripod.com, even though they ostensibly have very different content compiled by entirely different people? Should poetry.about.com be considered apart from pediatrics.about.com in spite of being managed by the same parent company?

Historically, at least dating back to the mid-Nineties, all of the search engines have treated subdomains (and, presumably by extension, sub-subdomains) as separate entities from their parent domain. I think there were good arguments for that decision, especially when a lot of free-site domains (tripod, geocities, angelfire, et all) were using subdomains for each of millions of free sites. Treating all of Tripod as simply one site would potentially "hide" a lot of relevant content from searchers. Similarly, even today, Universities on .edu TLDs still make extensive use of subdomains.

Personally, I believe subdomains (and, yea, sub-subdomains) should remain independent from each other and from the parent domain. That opens a door for potential abuse, of course, but the alternative (hiding relevant content) is far worse. And, geesh, these days just about anything opens a door for abuse after all.

One solution isn't to change the way subdomains are handled, but rather perhaps to fine-tune the Clustering algorithms to better recognize relationships between sites.

Google, I believe, has been trying to do that for some time now.

A friend told me an interesting story about two domains (not subdomains) he owned that were targeting the same keywords. One site ranked very well, but the other could rarely even be found in the same SERPs. A technical snafu went unnoticed long enough to cause the well ranking site to be dropped entirely by Google and, lo and behold, the second site suddenly started ranking extremely well for the same keywords which, just a month earlier, it couldn't buy a decent position. When the snafu was fixed and the other site indexed again, the second site once again slipped into search engine obscurity.

Clearly, I think, Google knew the two sites were closely related and maintained by the same person, and I believe their clustering algorithm refused to show both sites in the same SERPs. I've never been able to duplicate the situation (in large part because I'm reluctant to get a well ranking site dropped from the index!), but a few simple tests have convinced me Google determines the relationship by examining much more than just registrar data. I believe the threshold is likely set very high and Google only clusters related domains when it is VERY certain of the relationship. Not incidentally, clustering policy has changed in the past year, now applying to SERP pages instead of entire searches, so it's now possible to get a second related site on page two. Even with that, though, if I'm right, SERPs like those for 'queer forum' could probably be corrected by simply turning the dial a little and setting the relationship threshold a bit lower.

I'm not sure, however, doing so would be a good idea.

I suspect that ANY solution to listings like those we see for 'queer forum' would create more problems than it eliminates. Stopping SE spam, after all, is actually very, very simple ... don't return any results for any searches. Those kinds of solutions, however, ones that throw out too much relevant data with the bad, obviously aren't good solutions for anyone. The trick remains to eliminate bad results without affecting any good results. Google, I think, is already walking a very fine line. Their obsession with spam has resulted in much less spam, but I think it has also resulted in a less relevant search engine. Spurious SERPs, like those discussed here, are probably always going to be the cost incurred by trying to capture good, relevant data.

It should be noted, too, that the recent indexing of some five billion spam pages has absolutely nothing to do with subdomains or sub-subdomains. The real blackhats have been doing the same thing, albeit on smaller scales, with cheap domains. Using subdomains, essentially free instead of just cheap, broadened the scale while simultaneously making it MUCH easier to catch, but are certainly not an essential ingredient in the scheme.

Suppose I have a vacation property rental website with the Home Page at www.rentals7235.com. I want to put up a web page for a rental property at Mont Tremblant in Quebec. Which of the two following URL's would likely work better?

vacation.monttremblant.rentals7235.com

www.monttremblantrentals7235.com

Barry, if the search engines are to treat domains, subdomains, and sub-subdomains as separate entities (as they have traditionally done), it won't make any difference at all as far as rankings are concerned which you choose to use. Pages will be ranked via on-page and off-page factors, not by the domain name.

You should also know, by the way, that www.monttremblantrentals7235.com IS a subdomain, just as much as is monttremblant.rentals7235.com or vacation.monttremblant.rentals7235.com.

As you alluded, Barry, rumors continue to circulate that subdomains are a way to bypass the alleged sandbox/aging delay/"factors that could be perceived as a sandbox" phenomenon. I can't personally speak to that, however, as I've neither experienced the phenomenon nor tried to start a new site on a subdomain of an old site.

[send2paul]

This is a good subject Barry

I registered a domain a few months with a catchy name, for the sole purpose of creating subdomains - based on the accepted principle at the time, (as I read it), that subdomains where seen as separate websites.

So, up I go and set up a one page subdomain back on May 19th, and I submitted it to Yahoo, Google and MSN - just for good measure, of course

To date, Mr Google Bot or Mr Yahoo Bot has not visited yet, and Mr MSN Bot made a brief appearance on June 2nd. But! - All other visits to the page have been from referral links - none from search engine result searches. Of course, the content of the page, SEO etc could play a part as to why there are no search engine results, but the page is also on a subject which would be looked for by folks.

Does that throw some more confusion on the debate?

Paul

p.s. just read Ron's post - nice one

[webhostingrebates]

This is an excellent subject and one that I have a personal interest in.

I have registered a nice domain name that I am using explicitly for creating subdomains just like Paul. I have created several subdomains and put sites on them over the last 2 months. So far all of them have been indexed in the three big search engines and several of the sites have decent rankings (not great rankings due to the age of the TLD (4 months old))

All the sites target completely different markets and have good original content. The sites are set up like about.com sites in that they all link to a top level directory which then links to all the sites. So their is a definitely linkage between all the subdomains and the top level domain.

Charles

[yannis]

There were actually a few search engine weaknesses that were exploited. The main one was Google's policy of listing at least the index page of any domain. Total de-indexing is done manually (as far as I know).

There were a few domains with variations to t1ps2see, at1ps2see, bt1ps2see etc...

These domains were used to create subdomains and sub-sub-domains etc.

The main domain had no content on except an e-mail! I did actually send an e-mail but so far I have received no reply! I am hundred percent sure as I looked it up when the thread was active. (Lesson here if you are going to create millions of subdomains do not link them on the main domain to avoid penalties on the main domain!).

The success of the network depended on the sheer volume of pages. Their linking strategy did not involve the subdomains. As far as I could determine none of the sub-domains were linked together. Although they included some common spamming techniques like guest books. I do not want to reveal their main strategy for getting links as I don't want to encourage copycats. Any of the established members of these forums can pm me for it.

As for Barry's question are there are any lessons here? Apart for the obvious lesson that if you want to do black hat don't overdo it and that any black hat strategy is short-lived there are benefits. The main benefit here of creating a subdomain instead of a page is that the page will score better in the SERPS than an equivalent page in the main website i.e bluewidgets.mainsite.com will do better than mainsite.com/bluewidgets.html. There are other lessons here if one wants to do black hat adsense experiments but I will leave these for the more adventurous!

Anyway I take my white hat to this black hat Moldavian girl!

[Ruud]

Which of the two following URL's would likely work better?

The question is too vague, in my mind. It misses the what: which works better for ... ??

In your example the situation is basically identical. Both entities are standalone ones and from the URL you're suggesting the content page you want to put up is the default page served.

Is there a difference in example.com, subdomain.example.com, or either of those with a page in the URL?

It may be "harder" to get a page indexed on either vs. at least the one default page. But would those default pages be easier to rank?

[bwelford]

Thanks for the excellent and thought-provoking responses from all and particularly that winner from Ron. It all seems fairly clear although I did see a possible point of contention in what yannis said:

The main benefit here of creating a subdomain instead of a page is that the page will score better in the SERPS than an equivalent page in the main website i.e bluewidgets.mainsite.com will do better than mainsite.com/bluewidgets.html.

Ruud was coming at the same point too.

Again to clarify any difference of opinion do the following two web pages rank equally well or is one better than the other:

• the default web page, which has the URL bluewidgets.mainsite.com/ or
• the same default web page content named bluewidgets.html.and not treated as the default web page. So it's URL would be bluewidgets.mainsite.com/bluewidgets.html

Assuming there are no inlinks differences in the two situations (or back links to the Googlers), I assume that the two would rank the same. I believe yannis was saying something different.

[Ron Carnell]

In my opinion, Barry, the answer to your question will depend on the search engine. MSN gives much more weight to a keyword in the URL than does Yahoo, for example, while Google appears to give only marginal weight to a keyword in the URL.

In every case, however, bluewidgets.com is going to rank better than bluewidgets.mainsite.com (all other things being equal). In other words, a subdomain "might" get a boost over a primary domain, but ONLY because it's easier to stuff it with a keyword. It's the presence of the keyword that adds a slight boost, not the fact that it's a subdomain.

[Eddie]

Sub domains are a great way to manage themed sites without going to the trouble and expense of setting up seperate domains, and they do seem to be spidered well.

I've done this with my community sit about Gloucestershire. I have sub domains devoted to old photos, Gloucestershire transport history and a multimedia magazine. Of course they all link back to the main site.

[egain]

Would have to agree with you there Ron. Google at the end of the day reacts far differently to MSN/Yahoo which are IMO are far easier to forecast for.

[Aaron Pratt]

http://multiply.com/

do a little looking around at that, rent a blog, all have adsense on them by default, I find many that are only a few months old ranking with the best of them.

my own personal domains "subdomains" are doing very well out of the gate, tried the same thing on a newer one with no luck at all.

main domain vouches for the subdomain, if main domain is trusted subdomains are a great way to separate and focus content.

[yannis]

main domain vouches for the subdomain, if main domain is trusted subdomains are a great way to separate and focus content.

This is for sure. In addition linking between such sub-domains is viewed differently by the search engines (think Blogger and all the interlinking between blogs!).

But the main advantage of the spammers above was the sheer volume of sub-domains and the one page index rule, ensured that they were harvesting all the tail-end queries for a while. As more and more people clicked, the main domain started getting ranking from Google's user popularity routines.

Yannis

[Aaron Pratt]

I do not link back from subdomains to main domain because I believe it cancels the vote, think of it as a downward directory tree. Subdomains can survive on their own merit and will eventually get natural links and act as unique sites BUT in the beginning they need the link vote from the main domain to get crawled and indexed correctly.

[yannis]

Aaron

You are right not to link from the subdomains to the main domain. My point which I understand maybe it was not phrased properly is that linking between the subdomains is o.k. even from the same IP address i.e they must all be living on the same IP in any case and they will gain some pagerank. (imagine all the interlinking in blogger!). The intersting observation from the spammers main site was there was nothing there except an e-mail and a short message! It is still puzzling me!

Yannis

[Eddie]

You are right not to link from the subdomains to the main domain

Why ? A sub domain is just another site.

[Halfdeck]

One advantage of using subdomains or sub subdomains instead of /folder/.html I found is that Google indexes those subdomains far more frequently than deep pages on a single domain. That allows for gazillion pages getting indexed within a relatively short timespan. When I experimented with subdomains, I had around 1,000 subdomains getting freshly cached every day or every two days. Google would index 100 new subdomains in a matter of days. It's clearly a way to get around Google's trust filter.

From my experience, you don't rank any higher in the SERPS by using subdomains. 600 subdomains for me pulled around 1k uniques/day. That traffic level is identical/slightly less than when I had those pages under a single domain as /folder/. For the keywords I was targetting, position was roughly the same in both situations.

[yannis]

Halfdeck

Thanks for sharing with us your experience with subdomains. Did you have all those sub-domains under only one main domain or were they spread out over a number of main domains? Any interlinking?

One more advantage I would like to mention is that all those pages will be in the main index whereas as pages they may end up in the supplemental results.


Yannis

[Halfdeck]

Hi Yannis,

Sorry for the slow reply. I tried to gut my main computer a couple nights ago, and ended up frying my CPU.

Did you have all those sub-domains under only one main domain or were they spread out over a number of main domains?

I had the subdomains setup under one domain, ala blogspot.

Any interlinking?

Yes, the linking structure was set up as if each subdomains were a page belonging to the main domain. They all shared the same nav menu links, and every subdomain linked back to the main domain.

One more advantage I would like to mention is that all those pages will be in the main index whereas as pages they may end up in the supplemental results.

Agreed. The main reason I tried subdomains was to circumvent the supplemental listing problem.