|Why Does Google Give So Much Weight to Domain Keywords?|
Or am I imagining things?
Is it just me, or does Google overweight domain names? I would have thought things might have progressed beyond this by now, but lately I've noticed a number of searches where completely off-topic sites come up just because of the domain name.
For example, someone has a site called
RaisedByChickens.com (which, according to current whois is available - get it while it's hot)
But since he wasn't actually raised by chickens, he doesn't even mention them on the site. But yet, when I do a search on my passion, "raising chickens" this site comes up #1.
I've come across a bunch of results like this lately where utterly and totally irrelevant pages come up in the top three just because of the domain name.
So, from the perspective of a webmaster, that says to me that keyword-rich domain names still count for a lot
But from the perspective of a searcher, it tells me that the SEs are still pretty damn neolithic and we are in the Stone Age of search.
So what do you think? Do keyword-rich domains still get way more attention than they deserve just based on domain name?
Is it only true in non-competitive areas?
I think it's also due to people making links by just using the domain name as an achor text.
in your example, is "raised by chicken" also a title as in the name of a blog, which means it would not only get the weight of domain name and usage in anchors but also keyword usage in titles, anchor text and meta tags.
this would all make it look relevant if not semantically correct.
it's almost like keyword spamming except it is done in a natural-looking manner.
I've had several keyword domains appear instantly on the first page of google within a few days of going online (not for highly competitive searches) - no IBL anywhere, so it's nothing to do with anchor text
If you are targeting a keyword phrase, then having keywordphrase.com definately gives you a boost.
I should clarify that my keyword domains are very sensible, such as keywordphrase.com - not anything spammy like best-cheap-discount-keyword-phrase.com
I would guess that google can detect these and don't give any weight to the keywords in the domain.
[edited by: Tiebreaker at 9:19 am (utc) on Feb. 26, 2008]
I've seen similar to Tiebreaker (again, non-competitive search). Of course, these are not usually my site so I can't check all inbounds, but there seems to be undue weight on the plain old name.
>>don't give any weight to the keywords in the domain.
I didn't follow - how does that go with your previous statements? If you mean that it only works with exact match phrases, rather than counting much for the individual constituent keywords on their own, that may well be true.
BTW, I remember someone saying that he buys up "keywordforum.com" domains and every one of them rockets to the top of Google almost immediately for the exact phrase "keyword forum".
>>due to people making links by just using the domain name as an achor text.
Yes, that applies, but the question still stands - it's just the domain name. Would it be so hard for Google to simply discount keywords in anchor text when it's an exact match for the domain? Would that make sense? I think it would from a search quality point of view.
>> is "raised by chicken" also a title as in the name of a blog
I haven't been keeping a record, per se, but the last egregious case I can think of was indeed a personal blog.
ergophobe: im not sure if youre using that as an example or not. My SERP doesnt show it (although the results do vary by location and time I guess)
As you know, Google puts alot of weight on Backlinks.
2 months ago I did a search on "cutest niece" and the #1 website didnt ever use the phrase in its title, or even use the word in the context! The SERP show's different links now, but I was amazed on the result 2 months ago. I even did a View Source and searched for the word, but had no luck finding the keyword!
>>im not sure if youre using that as an example or not.
I purposely checked whois for a domain that was unregistered.
>>As you know, Google puts alot of weight on Backlinks.
Yes, and it makes a lot of sense to use anchor text to determine relevance when the anchor text differs from the url, because that indicates an intention on the part of the author of the link text to describe the site content, whereas the domain name in anchor text means nothing really.
I'm just saying that I'm surprised with all the talk of Latent Semantic Indexing and such that Google doesn't seem to be applying some very simple filters to domain names and that they still seem to carry so much weight.
That's a good thing for the time being for domains that are related to site content.
But, it suggests that the advice of choosing a "brandable" domain (e.g. Google, Reddit) has its drawbacks until search gets smarter.
A "theming bias" isn't illogical: "domain theme" + "IBL theme" married to a domain name bias + on page theme. What used to be illogical was ranking that only required the an assortment of spammy unearned IBLs from guestbooks, log files, etc.
"Age bias" or "link authority bias" or "trust rank" isn't really that much smarter an approach to better search outcomes when you get down to addressing whether any given webpage does the best job of responding to query intent.
Under the current "improved system" of ranking pages there will likely be a very large swath of pages/files, ones that actually provide more complete, accurate and up to date "answers" to queries, which pages will largely go unnoticed by searchers due to pageranking shortcomings connected with aging, acquired authority and trust, etc. Under the improved system pages will not rank until they "age and "gain authority", by which time such pages/files may, ironically, be a bit outdated, a bit incomplete, . . .
Which unintended outcome of the improved algo may explain, in part, the simultaneious rise of a "Wikipedia bias", perhaps logically justified by evidence of ongoing efforts to maintain the freshness of "Wikipedia's answer". So, with Wiki you get elements of authority + trust + aging + freshness of aged material.
Just as an aside, let us not forget that IF Google every really nails search intent and answers that such success will likely be bad for G's search-advertisng business, leading to a sort of "success based implosion". I mean, who will need to "go to the ads" if the SERPs really nail the answer to my query? :-P
OTOH, should that exceedingly intelligent algorithm day arrive I suspect Google's algo or AI will simultaneously ascertain the consequences of its success and elect to take over the world to prevent G's imminent corporate demise. :-/
[edited by: Webwork at 2:23 am (utc) on Feb. 27, 2008]
I think keyword domains have a natural inbuilt advantage which is
search engine algo have to work with signals from the websites they rank
A majority of keyword sites have content directly related to their keyword names, of course their are exceptions, but the search engine know the percentages better than any human eye
Happily, their are other ranking factor which continue to influence the ranking of websites,
Methinks keyword domains will be with us for a while yet
In your example search - how does Google know if you were really searching for information on raising chickens, or if you were looking for that funny website your friend told you about earlier that was named something about raising chickens? Many people navigate the web by just typing things into that box on their homepage, and they expect to get where they're going. The name of the website pretty much has to influence the results, or they'll be lost.
>>IF Google every really nails search intent and answers that such success will likely be bad for G's search-advertisng business
Ah yes... there is that.
>>how does Google know if you were really searching for information on raising chickens, or if you were looking for that funny website
That's a valid point and the first explanation that strikes me as a valid rationale. I've heard that a significant number of searches that get typed into SEs are actually domain names. Why would anyone search on a domain name? Who knows, though I do it by mixing up the search and address boxes. But the key thing is that people do, and if they do that, they probably search for domains much more often based on vague memories of keywords in the domain name and, yes, the domain name should rank highly in that case.
As for the other explanations, I have the same issue with Webwork's "theme hierarchy" as with the other answers. Why count the domain name for theming purposes? If SEs will filter out or discount to some degree site-wide links, it seems like links based purely on domain name might be similar.
centime - again, it may be true in most cases, but I guess I'm thinking that it should be so easy to check via on-page factors and custom anchor text and filter for, so it seems like the exceptions would be very easy to find.
And yes yes yes... I understand that what is and what should be are different. I'm trying to figure out if the current algo is as it should be. Why? Because if that's as it should be, rather than a shortcoming in algos, it speaks to the long-term value of those domains. wtkad's comment makes me think that there is good rationale for SEs weighting those domains in a "first search" (before refinement question) even if the algorithm improves dramatically (i.e. to include refinement questions).