Forum Moderators: open
Here is the proof. Do a search for allinurl:"web design" (or choose another popular two word phrase, enclosed in quotes)
If you can find a single search result that that includes "webdesign" (or the concatenation of your chosen keywords) in the URL, then, and only then, can we make the claim that Google is indeed parsing keywords out of URLs.
Do a search for allinur.. then, and only then can we make the claim Google is parsing keywords
I think the real proof would come from ranking for a nonsense word that only appeared concatenated in your domain.
Just ran the allinurl search, and happened to run across a site that had both "bluewidgets" and "blue-widgets" in the URL (domain and directory name).
Allinurl: "blue widgets"
highlighted only "blue-widgets" in the URL.
Allinurl: "blue" "widgets"
highlighted both "bluewidgets" and "blue-widgets"
1) Choose two madeup words. Check that they don't exist in the Google index.
2) Put the words together in an url.
3) Submit the url, use the toolbar, whatever, but don't link to it and don't place the words on the page.
4) Wait for the page to be indexed. Remember Google can index orphaned pages.
5) Type in the two made up words and see if Google finds the page.
6) Like everyone else who has done this experiment (I'm not one of them) keep the answer to yourself.
Kaled.
Suddenly I am seeing searches for "green widgets" highlighting URLs like this:
www.example.com/catalog_green_and_blue_widgets.html
That's good news for a site I built 5 years ago with all files separated by underscores.
For instance, a search for ialalp returns one page, and it is totally unrelated to the domnain I lifted the letters from.
This would support the idea that it is a display funciton only.
WBF
Choose two madeup words
Having done this, unless a positive result occurs, a negative result cannot be assumed until at least 2 updates have taken place, or, say, 10 weeks have passed.
Of course, you can run the second stage experiment concurrently with the first stage provided that different made-up words are used.
Kaled.
Why bother? If the Google use the keyword in URL as a factor in the algo, they sure aint putting much weight into it - from my experiences!
If you aren't doing good in the SERPS - it most likely has nothing to with keyword in the URL or not!
Example: What is Google? It's mainly a search engine... Does the URL contain the keywords search engine? No..
Does Google.com rank high for the quiry 'search engine'? Yes... (as we know, ironically they're not first in Google for that term)
well we have to discuss these silly ideas somewhere! BTW I'm being sarcastic!
Why bother with this thread? Why post if you don't care?
I have no particular interest in the inner workings of Google, but other people do. If there is an easy experiment that can be carried out I'll suggest it but I'm not about to bother conducting it myself.
Kaled.
Redicolous tread....
most people find if they sing the words they lose their lisp :-)
The point of the thread was the time old question of whether keywords in non hyphenated domains can be read by google. Google highlighting keywords in urls was new. You might not like the thread but it is not rediculous. There was no mention of its importance on ranking. However even if its significance was .01% of the algo or less then any SEO would be interested.
The thing I am excited about is seeing Google highlight keywords separated by underscores - it appeared that prior to this G considered KeyWord1_Keyword2 to be a single word and KeyWord1-Keyword2 and two words.
Suddenly I am seeing searches for "green widgets" highlighting URLs like this:www.example.com/catalog_green_and_blue_widgets.html
That's good news for a site I built 5 years ago with all files separated by underscores.
Shame it only does it for non-capitalised terms! That should be easy for Google to fix. *hint* :)
Google was one of the last to work out that company's often give themselves names that don't always describe what they do. In Britain 'Camelot' is the lottery gaming commission, so a search for this brings up lottery information rather than medieval chat rooms or whatever.
Obviously if Google gave weight to domain names they are using a very risky piece of information that isn't always telling the full story.
Recent results have shown that Google gives more weight to keywords in file names rather than domain names but I expect now that we know this they will lower the weighting given to that aswell ...
Thanks
Harwich
the next stage is to introduce those words into the body text of other pages. In this way, the made-up words will be introduced into the Google dictionary
Now we're getting somewhere. In the small bit of research I've done, you would have to have the term on a ton pages to get the term introduced into the "dictionary". I've been looking at some domain fragments, and they are only bolded when the fragment appears in the body of the page w/ the concatenated url.
Thanks goes out to whomever is making thousands of pages of only domain typos. It was the only place I could find url fragments in the text of the page
In the small bit of research I've done, you would have to have the term on a ton pages to get the term introduced into the "dictionary".
I haven't researched this, however, I placed a madeup word on a critical page of my site when first published simply so that I could tell if it had been indexed (by searching for it). This worked perfectly suggesting that Google's dictionary only requires the word to appear on a single page. However, it is possible that two dictionaries operate, and that a word has to exist on perhaps thousands of pages before it enters the second dictionary and only words in that second dictionary can be used usefully (non-hyphenated) in urls.
I can see no reason for operating a two-dictionary system, but perhaps if I gave the matter more thought I might come up with one.
Kaled.
I placed a madeup word on a critical page of my site
I can see no reason for operating a two-dictionary system, but perhaps if I gave the matter more thought I might come up with one.
If you think logically, a company will generally buy a domain with their company name in it which may have nothing to do with their services (like the name of a solicitors firm), the only people who will generally buy a domain name full of keywords are spammers and affiliate marketers. Therefore only a 'moron' would use the domain name as a ranking feature anymore.
If you don't think widgets.com/red/ is more likely to be about red widgets than shshts.com/tsref/ then that's your choice. But I'm going to consider that an extremely bad one.
And what does a company have to do with it? We are talking about websites. If your search engine doesn't seek to rank web pages or websites, that speaks for itself.
If you don't think widgets.com/red/ is more likely to be about red widgets than shshts.com/tsref/ then that's your choice. But I'm going to consider that an extremely bad one.
Grow up. It think that widgets.com/red/ that doesn't mention red widgets is less important than shshts.com/tsref that does. And if both were to mention red widgets then it would turn to anchor text, themes and so on.
These days the chances are the widgets.com/red/ is an affiliate spam site pointing to shshts.com/tsref/
And what does a company have to do with it? We are talking about websites. If your search engine doesn't seek to rank websites, that speaks for itself.
You do realise how ridiculous that sounds don't you?
It think that widgets.com/red/ that doesn't mention red widgets is less important than shshts.com/tsref that does.
That's a very sensible and correct statement.
Google is looking at what your site represents and not what it's called. Other SE's should (and in the main are) doing the same.
However, from a user perspective, I agree with SteveB. The former weidgets.com/red/ is going to look far more enticing to a user looking at a page of results than shts.com/tsref.
TJ
Just to keep the algo simple, I imagine keywords in domain names are equally valid, however, the argument that domain names are often unrelated to products is a good one and Google may take the same view.
The issue of highlighting is interesting, but it may simply be that the highlighting algo it not word-orientated. It may simply highlight string occurrences. In any case, the highlighting algo is likely to be unconnected with search/filter algos.
Kaled.
