SEO Test

Discussion in 'Search Engine Optimization' started by Jim_Westergren, Aug 11, 2005.

  1. #1
    Jim_Westergren, Aug 11, 2005 IP
    Design Agent and Epica like this.
  2. bigdoug

    bigdoug Peon

    Messages:
    845
    Likes Received:
    54
    Best Answers:
    0
    Trophy Points:
    0
    #2
    Keep up the good work, it looks like fun:)

    D
     
    bigdoug, Aug 11, 2005 IP
  3. Jim_Westergren

    Jim_Westergren Notable Member

    Messages:
    1,882
    Likes Received:
    247
    Best Answers:
    0
    Trophy Points:
    235
    #3
    Small update for you guys:

    Google has now listed 7 test pages. 6 of 16 from test number 1 and 1 of test number 2.

    Unfortunately it hit the filter as "similar pages" but I have remedied that now by adding more unique random Lorem Ipsum text to both title and body tag.

    But for right now Google has listed:

    #1: KW in domain
    #2: KW as anchartext
    #3: KW in title
    #4: KW in URL
    #5: KW spammed in meta tags (15 X)
    #6: KW in meta tags

    As for the second test about the URLs Google has listed the www.domain.com/word1_word2_word3 version and not the other two ones yet.

    This already rises some interesting questions, meta tags maybe has a value after all? And why was not the other pages listed?

    Please note that this is only the beginning of the test and a complete analysis will be issued when it will be finished. MSN and Y has not yet indexed any pages yet and actually I would need more links to the SEO test page to speed of the listings of the pages, I could give links back, yes.

    Will keep you updated as I assume there can be some interest about this on DP.

    Feedback/suggestions is appreciated.

    Thank you.
     
    Jim_Westergren, Aug 12, 2005 IP
    web-rover likes this.
  4. fryman

    fryman Kiss my rep

    Messages:
    9,604
    Likes Received:
    777
    Best Answers:
    0
    Trophy Points:
    370
    #4
    I get a 404 at the test page, could you check the url?
     
    fryman, Aug 12, 2005 IP
  5. Jim_Westergren

    Jim_Westergren Notable Member

    Messages:
    1,882
    Likes Received:
    247
    Best Answers:
    0
    Trophy Points:
    235
    #5
    I just changed web host account, it is now working again.
     
    Jim_Westergren, Aug 13, 2005 IP
  6. namxas

    namxas Peon

    Messages:
    42
    Likes Received:
    0
    Best Answers:
    0
    Trophy Points:
    0
    #6
    good work, thanks
     
    namxas, Aug 13, 2005 IP
  7. emz

    emz Peon

    Messages:
    364
    Likes Received:
    10
    Best Answers:
    0
    Trophy Points:
    0
    #7
    excellent idea, looks like a great test.
     
    emz, Aug 13, 2005 IP
  8. web-rover

    web-rover Peon

    Messages:
    1,341
    Likes Received:
    45
    Best Answers:
    0
    Trophy Points:
    0
    #8
    cool beans

    keep us updated
     
    web-rover, Aug 13, 2005 IP
  9. Jim_Westergren

    Jim_Westergren Notable Member

    Messages:
    1,882
    Likes Received:
    247
    Best Answers:
    0
    Trophy Points:
    235
    #9
    Sure. The testpages needs to be updated in the SE listings as I had to change them and make them more unique. This will take 1-3 days.

    As of my second post on this thread with the standings in Google this is not accurate and should not be looked at. Google saw the pages as similar and the standings were done by using the "site:jw-webbdesign.se KW" command. But I have realized recently that when you do the "site:" command in Google it most probably listes the sites total random and not in any kind of ranking.

    But this factor is handled and in 1-3 days we will have the pages unique in the listings.

    But one thing I can say now, the page Google is currently ranking number one is one of the 5 pages that is on a different domain. And which domain? The only one that is DMOZ listed ... (not the highest PR ...)

    Actually do any of you want to test something else? I can make test number 3 easily, just let me know and I do it (except with domains).

    And could anyone provide a link to the test page? I will give a link back (maybe even better) to you but it is needed to speed up the test.
     
    Jim_Westergren, Aug 14, 2005 IP
  10. Key

    Key Peon

    Messages:
    40
    Likes Received:
    0
    Best Answers:
    0
    Trophy Points:
    0
    #10
    This looks like excellent testing procedure. Glad to see you going about this thoroughly!
     
    Key, Aug 14, 2005 IP
  11. Jim_Westergren

    Jim_Westergren Notable Member

    Messages:
    1,882
    Likes Received:
    247
    Best Answers:
    0
    Trophy Points:
    235
    #11
    Yes, thank you.

    I have studied earlier SEO tests and their mistake is that their pages were not unique and that they had all the test pages on the same domain.

    With LIpsum.com I can generate unique random text on the exact byte of size and that tool is the key to these tests.

    It will be very interesting to see the results.
     
    Jim_Westergren, Aug 14, 2005 IP
  12. Jim_Westergren

    Jim_Westergren Notable Member

    Messages:
    1,882
    Likes Received:
    247
    Best Answers:
    0
    Trophy Points:
    235
    #12
    Here is the latest update (day 5):

    SEO Test number 1:
    Ranking on different variables

    Google rankings:

    #1 KW in title but on a different domain (PR 4, DMOZ listed)
    #2 KW in domain [EDIT: My own experience shows that new domains rank very well for some weeks before going into the sandbox]
    #3 KW in title but on a different domain (PR 4)
    #4 KW in anchortext
    #5 KW in title
    #6 KW in body text 5 times extra
    #7 KW in invisible text in body (5 X)
    #8 KW in directory name
    #9 KW in URL
    #10 KW in meta tags
    #11 KW in H1
    #12 KW in H2
    #13 KW in title but on a different domain (PR not updated, probably 3-4)
    #14 KW in title but on a different domain (PR 4)

    MSN rankings:

    #1 KW in title but on a different domain (PR 4)
    #2 KW in title but on a different domain (PR 4)
    #3 KW in title but on a different domain (PR not updated, probably 3-4)
    #4 KW in title but on a different domain (PR N/A)
    #5 KW in title
    #6 KW in body text 5 times extra
    #7 KW in invisible text in body (5 X)
    #8 KW in meta tags
    #9 KW in anchortext
    #10 KW in domain [EDIT: MSN sandbox ??]

    Yahoo has not listed any yet.

    Conclusions (as per the test findings so far):

    1. Google puts much emphasize on DMOZ listings.
    2. Google consider KW in domain more valuable than in the title.
    3. Google puts much emphasize on the anchortexts of the links.
    4. Google doesn't penalize invisible text.
    5. Meta tags DOES have a value in both Google and MSN.
    6. H1 and H2 has been overestimated.
    7. MSN puts much emphasize on titles.
    8. MSN does not value KW in domain so much.
    9. MSN does not value KW in anchortext so much. (If not thousands of links, like huge sitewide ones in which this variable overrules the other ones.)

    SEO Test number 2:
    What is the best way to write the URLs?

    - Google prefers first "-", then "+", then without anything and then "_".
    - MSN prefers first "_", then "-", then without anything and has not listed the "+" version.

    Conclusion:

    Write your URLs by using the "-" between the words as this is best for Google and almost the best for MSN. With Yahoo we will wait and see.

    I'll post here again if I see some major changes but I think it will keep like this. I also miss the Yahoo data and will provide it as soon as I get it, as well data from new tests I will do.
     
    Jim_Westergren, Aug 15, 2005 IP
  13. Googles76

    Googles76 Peon

    Messages:
    664
    Likes Received:
    26
    Best Answers:
    0
    Trophy Points:
    0
    #13
    Hey, nice tests. Although I have a problem with this one:

    Everyone here knows what the coop can do to MSN rankings:D And that is just it, coop gives lots of anchortext with your desired keyword. Maybe I am just not understanding the test:confused:
     
    Googles76, Aug 15, 2005 IP
  14. Jim_Westergren

    Jim_Westergren Notable Member

    Messages:
    1,882
    Likes Received:
    247
    Best Answers:
    0
    Trophy Points:
    235
    #14
    You are right.

    I need reformulate that line. I have myself seen what anchor text can do for MSN but in what I saw it was on sitewide links on 10.000+ pages.

    So I think in the MSN algo that you get that much power value providing it is a huge number of links with that/those KWs in the anchor text while if you compare 1 link with anchor text with Google, the one from Google is stronger. But Google's algo is probably better at the anchor text factor so that you can't "Google bomb" easily.

    You see what I mean?

    I have now edited that sentence.
     
    Jim_Westergren, Aug 15, 2005 IP