WARNING! READ THIS FIRST! WARNING! READ THIS FIRST!

This page exists to run experiments on how search engines rank pages. In order for an experiment to be effective it needs to be run in a controlled environment. That means that all variables need to be controlled. Since incoming links are a variable that search engines use to rank pages it is vitally important that we control the incoming links for this site.

You MAY link to THIS PAGE and only this page. You MAY NOT link to ANY OTHER PAGE on this site. We do not grant permission for anyone to link to any pages on this site other than the homepage. If you violate this policy we will brand you as someone who is either illiterate, stupid, or wants to invalidate these experiments for whatever reason. Your name will be placed on our wall of shame for all to see.

I have created this page in order to obtain empircal evidence of search engine ranking algorithms. There is too much second guessing in the SEO field and too many incorrect conclusions. This site aims to fix that. Please note that you will not be able to see the results of these experiments until this site (launched 6/7/2003) is listed in search engines.

Expirements:

If you have ideas for other experience let me know. Email me at chris [at] searchenginelabs.com.

Do meta tags help your site get listed?

This is a very simple expirement and needs only 1 page for testing. The page in this test has a word in it's meta tags that does not exist elsewhere on the page or in the anchor text. So the only place the word exists is in the meta tags. I will not even state the word here since this page does link to that one and doing so could introduce another uncontrolled variable into the mix. This experiment is not testing whether or not meta tags are used to generate your listing in the SERPs (we know the description tag is sometimes used), we are only testing if they will get you listed on SERPs.

Do meta tags help your site's ranking?

This test is more complicated than the last one. Here we are testing if meta tags, in addition to body text, will help your site's ranking. See some people claim that meta tags only help when they contain the same words that are in the body, so using the easy listing test above does not work to settle the issue. There are 10 pages in this test, each contains the exact same keyword density, each contains only a different adjective in the body text to prevent the search engines from thinking that it is duplicate content. 5 of the pages use meta tags, 5 do not. If all 5 of the pages with meta tags rank higher than those without meta tags then the search engine in question uses meta tags to rank your site when they concur with the body content. Pages with even numbers in their filenames use meta tags, those with odd numbers do not. Searching for the word dog on any search engine while limiting results to this domain will show you the results of the test.

Date 9/25/2003 - Someone has questioned the validity of this test. Claiming that site specific searches use a different algorithm and that unfiltered searches (normal searches that allow more than two results from a single domain) also use a different algorithm. You can view an unfiltered search for this experiment here. The problem is in a normal Google search only 2 results from a single domain can be listed, and Google chose two meta tag pages. Of course since there was a 25% chance of that happening even if Google doesn't use meta tags that is hardly statistically significant. If you expand the search the first two are the same (yet some claim it is a different ranking algorithm) and the rest of the pages do not follow any pattern. In other words the meta tag pages are not all ranked above the non-meta tag pages. Yet some people claim that only default Google searches count and that everything else uses a different algorithm.

I have decided to swap all of the pages. So those that used to use meta tags have now been switched with those that used to not use meta tags. We will see what results this produces.

  • The Red Dog(meta)
  • The Orange Dog
  • The Yellow Dog(meta)
  • The Green Dog
  • The Blue Dog(meta)
  • The Indigo Dog
  • The Violet Dog(meta)
  • The Pink Dog
  • The Brown Dog(meta)
  • The Black Dog

    • Is it possible to be listed under a word thanks only to the anchor text of incoming links?

    This is another very simple test. We will link to a page using a word that does not exist on that page. We will then search on a search engine for that word while limiting results to this domain and if the page shows up then you can be listed thanks only to anchor text. However this experiment will not be conclusive if negative because we are unable to test the possibility of a very very weighty link. There may be a cutoff where lower weight links do not allow you to be listed but higher weight links, or multiple links, do.

    Do outgoing links increase your site's ranking?

    Personally I think that the notion of outgoing links helping your ranking is absurd and counter-intuitive. If all a webmaster had to do was to link to a high ranking site and then they could leech off that site then relevancy would decrease dramatically. Counting incoming links is good because it is a way to see what other people think of you. Counting outgoing links would only be asking what do you think of yourself -- the same thing meta tags did and we all know what happened to meta tags. There are search engines that use outgoing links, such as Teoma. However Teoma uses them to generate a secondary set of results called "Resources." That is a far cry from using them to generate actual SERPs. Additionally its kinda a step backward from the search engine perspective. The goal of a search engine is to point a user to a place where they can find the information they need. Pointing a user to a "hub" site that contains links to the correct information instead of pointing the person directly to the information does not make for a better user experience. If the search engine knows about the correct authority page why would it send someone to a hub page? Especially since the SERP is in effect, a hub page itself.

    But I digress. This test is very similar to the meta tag ranking test above. The difference is in this test there are multiple sets of 3 pages each. One page has a keyword rich outgoing link on it to an on topic site, one has a non-keyword outgoing link, and one has neither. However overall keyword density for the three pages is equal. To view results search for the set name, such as "A Tale of Two Cities", while limiting results to this domain. If both link pages are listed above the no-link page then outgoing links help. If only the keyword rich link is listed above then text in anchor tags is simply weighted more than body text but where the link points does not matter. If no page type is always on top of any other then outgoing links do not matter and anchor text is weighted the same as body text. If the no-link pages are on top then anchor text is weighted less than body text.

    Set 1: A Tale of Two Cities

    Set 2: Call of the Wild Set 3: The Three Musketeers

    Copyright © 2000-2003 Jalic LLC. All rights reserved.