I think I remember reading in WW that www.keyword1-keyword2-...-keywordn.html could be seen as keyword stuffing and potentially penalized by Google. Does anybody have more recent information on this, e.g. what number of keywords is over the edge? Better to avoid more than one hypen as a general rule?
kiril that is not true. Keyword stuffing is content-based not url based.
>kiril that is not true. Keyword stuffing is content-based not url based.
Correct. There might not be a penalty, but a reducing of the benefit of keywords in domains beyond a certain point. I'd say up to 2 hyphens is safe. This many are commonly found in non-spammy domains. I presume the advantage to less hyphens is that whatever keyword in domain benefit exists is divided by number of keywords. IOW keyword.tld gets 100% of the benefit for keyword. However with keyword-keyword2-keyword3.tld, just "keyword" gets 33.3% of the algo benefit. Of course, if people actually will be searching using "keyword keyword2 keyword3" a lot, then keyword-keyword2-keyword3.tld is ideal.
|>Can Google read the keyword in www.brandkeyword.com? |
|On a search for "keywordinfo" I got serps dominated by "keyword info", even highlighted in cache pages. |
I am intrigued. I don't see how we can be getting such different experiences. Those two statements will keep me awake tonight. :-)
It the same experience.
pageoneresults has a good point about the Yahoo SERPs.
And try those searches in your stickymail :)
|IOW keyword.tld gets 100% of the benefit for keyword. However with keyword-keyword2-keyword3.tld, just "keyword" gets 33.3% of the algo benefit. |
While domain names is more my field (and I've never known a search engine to parse non hypened words out of a domain name), and I'm 'way-far' from knowing alot about search engines, I was just today jotting down some notes and I have noticed the following about google-
1. If I search on "Jobs for word1 word2', the listings that come up are the ones that have JOBS listed first (or near first) in their title/content, followed by ones that have jobs listed later in their title/content, and so on.
2. The listings that had word1 close to being their second found word (just like I asked for in my search phrase) in their title/content were listed higher than ones with it later in their title/content.
3. The listings that had word2 closest to being the third word in their title/content were listed higher.
4. If the FIRST word repeated again in the title (Like "WE HAVE JOBS. THESE JOBS ARE FOR word1 word2"), the repeated word was STILL considered a 'first word', so the second 'JOBS', above, is only 3 words away from the word 'word1', not five - they don't seem to count from the first one in these cases.
5. So, the closer ALL the words were together (that matched my search phrase) in the title/content of the listing, and the closer they were to the beginning of the listing, the higher they were listed.
So in the example above, if someone had the exact match to JOBS FOR word1 word2 in the beginning of their title/content, they were listed nearer the top, above, say, a listing like- JOBS. WINKY DINKY IS HIRING WAITRESSES. WE ALSO NEED A word1 <word2 but not plural>.
See what I mean? In the second example, JOBS (the first word in my search) is 10 words away from my next searched word, word1, so it is listed lower.
Is this old news? There's obviously a correlation in there somewhere.
As for brandkeyword.com versus brand-keyword.com. The hypened one is definately better for the search engines, while the other is better for the lazy public (one less keystroke). So you really should have both!
Of course, in the end its what you have on the site that counts. If you build it well, they will come... is my theory.
[note- obviously I dont have that whole keyword thing fleshed out, but there IS something like that going on]
[edited by: ciml at 10:07 am (utc) on Oct. 1, 2003]
[edit reason] Anonimised keywords. [/edit]
> Look at adwords, play about with safe search.
I understand now. NFFC is wise...
If I searched for something keyword, the keyword would be bolded in the URI...
Just because the word is bolded, doesn't mean it counts in the algo. It could just be a regex in the output code eg.
Do they read keyword? Yes
Do they highlight keyword? Yes.
Do they parse it for ranking purposes? Not yet.
Check (arguable, just to grasp the idea):
allinurl:keyword info -keywordinfo
Highlighting is a completely different process!
I always try to buy both versions of the URL For Trademark and branding reasons. I have noticed that domains with dashes do best!
>I think I remember reading in WW that www.keyword1-keyword2-...-keywordn.html could be seen as keyword stuffing and potentially penalized by Google.
>Does anybody have more recent information on this, e.g. what number of keywords is over the edge? Better to avoid more than one hypen as a general rule?
I have two domains with more than two hyphens.
has been treated very badly - no backlinks and no PR. It does show up in the index if you actually search for it by name but it gets no traffic at all from Google. All other search engines send it hundreds of searches per day.
The other many-hyphen-domain-name-conglomeration.com domain name with more than two hyphens in it actually has five of them. I bought that as a joke and don't care if google ever does anything with it.
It also has no backlinks and no PR.
No other domains have been treated that way.
I have four or five two hyphen names and google treats them just fine, PR 5 and PR 6 and hundreds fo backlinks counted.
After eight months and about 600 pages of real, meaningful and compelling content added to the widgey-widget-widgey-widgets.com site, it still has no PR0 and no backlinks.
When I asked why, firstname.lastname@example.org replied first with
"It is in the index."
When I asked why it was PR0 and had no backlinks for eight months, they replied that they do not comment on such things.
I poked around a bit more and found that this page:
( as compared to widgey-widget-widgey-widgets.com )
has hundreds of backlinks and PR5 so it is not a case of "too many hyphens." There must be something else wrong with the widgey-widget-widgey-widgets.com page. Curiouser and curiouser - to have one page out of hundreds of thousands penalized is very strange indeed. Since the designs are similar, it must be a manual penalty, and a highly secretive one, at that.
To go back up to the original question, my $.02 answer is Yes - keywords in the domain do matter, hyphens or not. Google will pick up on two words in the domain based on anchor text, and the keywords title tag (among other factors).
Listed below are the key elements:
1) Keyword domain, hyphenated or not
2) Exact keyword phrase in the title, with phrase broken out ("brand keyword")
3) Exact keyword phrase in anchor text, phrase broken out
The title, anchor text, and other on-the-page factors help Google determine how the domain string should be broken up. This is Google's form of "linguistic intelligence", as referred to earlier in the thread.
I belive HitProf has hit the nail on the head.
The major assumption in pageoneresults' analysis is that bolded = counted for ranking purposes. That is a huge assumption (as toolkit also pointed out).
The current algo will not return www.brandkeyword.com for a query of "brand", unless other factors come into play (titles, anchor text, etc.) If someone can come up with a single SERP to prove otherwise, please do so.
I will offer the following SERP in support of my belief:
Note that this search for asterworl does not return www.webmasterworld.com.
|The major assumption in pageoneresults' analysis is that bolded = counted for ranking purposes. That is a huge assumption. |
I wouldn't say that I was assuming with my reply. I was stating fact based on research. Keywords in URI's are calculated into the relevance factor of a page. Don't get me wrong, as you pointed out, there must be other factors involved which there most definitely are.
Typically when you see a keyword domain, the overall theme of the site centers around that keyword or those keywords. I never stated that having just keyword.com would automatically put you in the top positions, that is not 100% correct. The only time that may apply would be in non-competitive instances.
There are many factors involved and the URI is just one of them. From my perspective, it could end up being one of the top ten things why a certain site ranks highly.
If you take two sites, exactly the same in content with all other factors being equal except the URI, the site with keyworddomain.com will outrank the site without keyworddomain.com 9 out of 10 times.
P.S. In response to your search query, try this one...
#5 out of 60,400,000 not bad at all. Of course that search was just as crazy as yours! ;)
[edited by: pageoneresults at 7:28 pm (utc) on Oct. 1, 2003]
>The major assumption in pageoneresults' analysis is that bolded = counted for ranking purposes
I've re-read the post, can't see anywhere that was stated. Part one of the original question was "Can Google read the keyword in www.brandkeyword.com?", if it counts in the algo is another question [and thread imho] althogether.
>Note that this search for asterworl does not return www.webmasterworld.com
asterworl is not a real word.
|asterworl is not a real word. |
It could easily be a brand name. What is the denifition of a real word anyway? Are you suggesting that Google scans every string of characters for a match against a database of known "real" words? It seems like they would be unable to do something like that in a scalable way.
#5 out of 60,400,000 not bad at all. Of course that search was just as crazy as yours! ;)
pageoneresults, that's an entiely different query, which proves nothing. For one, the world "webmaster" appears on the page. An even larger factor is likely the anchor text used to link to the page. Try this SERP:
The original question, as I understand it, was a very specific one: "Can Google read the keyword in www.brandkeyword.com?" Now the word "read" here is subjective, but, given the context of this forum, I think it safe to assume that "read" means roughly "counted for ranking purposes". Still no one has submitted a single SERP to prove that this is possible. To do so, of course, the page cannot have "brand" anywhere on the page, and cannot have "brand" in any anchor text. Is this realistic for a site targeting the word "brand"? Of course not - but the question is more academic than anything else.
>What is the denifition of a real word anyway?
In this case the answer might be here [18.104.22.168].
>Still no one has submitted a single SERP to prove that this is possible.
Sometimes you just have to trust. ;)
|In this case the answer might be here. |
Interesting theory. I've read that - and I won't pretend that I understood every single bit of it :-)
Yes, it does say that there exists a lexcion of words. It does not say anywhere, to best of my recollection, that the system looks for lexicon words within stringsoftextthatarenotseparatedbywordseparators. It's one thing to parse a page of content into words by using word separators like space, hyphen, etc. It seems like it would be exponentially more compuationally expensive to parse out words where there are no word separators to use as guides.
And what would be the benefit to doing that anyway with respect to page ranking? Seems like it might lead to poorer SERPs. A small example: "automatically" matches "auto", "mat", "call", "automatic", "ally", "iCal", and perhaps other "words". But how many of these "subwords" are relevant to ranking for the word "automatically"?
You're right about one thing, of course. Sometimes you do have to just trust :-) But in this case, there isn't even one piece of evidence that Google counts substrings as words for page ranking purposes. Sometimes you have to weigh the evidence, too :-)
>It's one thing to parse a page of content into words
Who said anything about pages, domains was the question. Old data from 2001 says that the median length was 11 chars [source: [zooknic.com...] ]
>A small example: "automatically" matches
Depends how you match, why not start left and end once you hit the longest word in the core lexicon, then repeat from there?
>evidence that Google counts substrings as words for page ranking purposes.
The question was "Can Google read the keyword in www.brandkeyword.com?", I say Yes.
I totally disagree. I've never known any search engine to read/pull a keyword out of an unbroken string of characters.
This could probably be tested with a spare domain name.
I pointed an extra domain of mine to a webmasterworld forum a few days ago. Would this be a valid test?
You could easily create a dictionary from the page contents,
and then hunt through the url for the words on the page.
Doh. Youre right. It is easy. I just proved it with my own domain.
site:webmasterworld.com world -qwerrew
Compared to this:
site:webmasterworld.com webmaster -qwerrew
site:webmasterworld.com webmasterworld -qwerrew
The domain keywords are never hightlighted except in the last example.
This is, of course, if highlighting matters in this particular case or if it is a totally different dog.
Thanks for the many responses. My site is #3 for my category keyword in 7,550,000 results and I am trying to get to the top. Originally, I thought that I had an edge by having my site name as www.brandkeyword.com, but I am now convinced this is not so.
My analysis of the top 20 sites for this keyword has nevertheless convinced me that www.brand-keyword.com is a huge advantage. It seems to be almost as powerful as www.keyword.com, although not as brandable. I am however stuck with my original URL so I have to work with what I have.
After trying all the SEO strategies I have learnt, I am now convinced that the only way I will get to the #1 position is by working even more assiduously on my link campaign.
I have a domain that is made up from 3 keywords like: www.keyword1keyword2keyword3.com it has always ranked #1:3 when searching on that key word phrase despite some sites below mine having ALL other factors in their favour. Most say that Google does not use this simply because the domain is not bolded when the term is used. As far as I know Google NEVER bolds domains, perhaps because it would give away the fact they are used?
My findings are the opposite in that Google does look in the domain for matching keywords. Nearly all programming languages can extract words from strings so I can see no reason why Google wouldn't.
Nobody is going to use the domain bluewidgets.com if they don't sell them.
Dave_Hawley, might this be due to lots of people linking to you with the anchor text "keyword1 keyword2 keyword3"? At the moment anchor text trumps all. 3 keywords phrases usually aren't that competitive, and you are doing exactly what would optimize anchor text links. What I'm suggesting in your specific case you'd be #1 with or without hyphens.
That is quite likley true. But links to this domian (only about 20 or less) also use Keyword1Keyword2Keyword3 which would suggest the Google can and does parse out the keywords from the one word. If it does this for anchor text, I can see no reason why Google wouldn't do the same for domains?
Im with Dave on this one.
I talked to Matt Cutts on Pubcon and asked why they do not allow wildcard searches [webmasterworld.com]
In other words, variations of parsing /stemming, but also being able to search for a part number of which one digit is missing.
More or less, his answer was; who says we are not doing/experimenting with things to that effect within normal searches (amongst others in Froogle)?
Personally I am not seeing very much of it at all.
Trying to distill "brand" and "keyword" from brandkeyword should technically not be very difficult though.
As others suggested, real words extraction, occuring e.g. on the webpage itself, or in the anchortexts towards it could make things quite simple, avoiding "asterworl" extractions from www.webmasterworld.com.
i have a non-keyword domain...it's made of two very common words. However on my site these two words only appear as my dommain, so word1word2. They are not seen on my site as word1 word2. No links point to me with word1 word2, only with them together with no seperation. A search for the words seperated as word1 word2 find my site at number one. This could not happen unless the links were being seperated to read word1 word2.
I got smacked on the head with confusion when I saw this in the results.
get this weirdness:
key1-key2-key3 as a query (with those dashes)
key1key2key3.com in the SERP
key1 key2 key3 as a query
only need to see it happen once to be certain google knows how to highlight in the middle of a concatenated url string - the fact that I can only reproduce it with a certain set of key1-key2-key3, and only with dashes, is all just fodder for confusion, but besides the point.
| This 63 message thread spans 3 pages: < < 63 ( 1  3 ) > > |