SEO4FUN: SEO Myth or Cold Hard FACT?
DISCLAIMER: I'm no professional SEO; like the domain name says, it's all about SEO for fun. I'm using this domain to run a few experiments to see how different setups effect ranking on Google, Yahoo, and MSN. I'll be writing most of my notes in my SEO Blog If you're looking for a free SEO course, check out that link.
Only First Link Counts Test
Here are two links to my blog home page using two anchor texts: Velvet Underground and Blondie. None of those words appear on any pages of the site except this page. So you should see two pages returned in search results for those terms: this page and blog index. (set up July 8th)
SEO Test 1
This is a test to measure the weight of various on page / off-page SE optimization methods. Note: Tests results should be checked on different DCs. If there's a difference in the SERP, then its a sign that the algo has been tweaked/changed. I'd also like to test pages with 3-4 word keyphrases, not just 2 word combos. I may have to modify the test pages so the keyword densities are nearly identical.
Note: Each page has a unique <title> and the on-page text starts off with unique words on every page to avoid dup filter. Not every page has META description tag.
How to Test Nofollow
The question for me is does nofollow really prevent PR leakage? Since I'll probably be dealing with PR 0~1, I'll need to build similar pages with an embedded keyword and compare one page with 100 links with nofollow on all of them versus another page with 100 links that doesn't use nofollow(?) Note that the PR of the page itself won't change. The difference we want to see is the PR of the pages this page links to. If one page gets a link from a non-nofollow 100 link page, then say it gets 0.01PR transferred. Another page with 99 nofollows and 1 clean link should get 0.02PR, so ranks higher in the SERP. As contrast, say I have another page that gets a nofollow link. That should get 0.00 PR and rank the lowest.
So how many pages do I need for this test? Say 2 pages with 50 links on each page. The links go to big sites like cnn.com (so I don't have to build crap pages). Page 1 will use nofollow on all of its links. Page 2 will not use nofollow. Now we also build page A that gets linked from Page 1 using a clean link, and page B that gets linked from page 2. Finally, we have page C that gets linked from page 1 with a nofollow on the link. As for content, title, meta tags, pages A, B, C should be nearly identical.
The keyword for this test is testxzrelnofollow seo4funf5s. The linkfarm pages does not contain keywords, just links to major sites.
Nofollow Page This page is populated by miscellaneous outbound links. PR/number of Incoming links to this page should be exactly identical to the next page.
Here's a list of things I observed from the tests so far. This is a list of DEFINITES. I should also double check this with other domains I already have listed. I'll move this section off to a separate page once there's enough content.
Google SEO Test Result
3/2/2006 Today I noticed Google indexed a few more pages from my domain. So I'm going to see how the pages are ranking and make a few notes here. The DC I'm checking. Note though, that some listings might be TIED in position. Interestingly, so far I haven't seen any supplementals, at least not for those pages where supplementals were forced.
- Brandnewsacx1 sasahz returns 10 results.
- http://www.seo4fun.com/test-1/brandnewsacx1-sasahz.html came up first, which combined h1 and other stuff. This page was also linked from the index with keyword in link text.
- 2: Text at the beginning of title. Note that the text also occurs once on the page.
- 3. Keyword in title and bold on page.
- 4. Unoptimized page ??? This page has nothing in title, metas, h1. Just one occurance of keyword on the page. I don't understand how this page would rank higher than other pages.
- 5. keyword in H1.
- 6. keyword in h2.
- large block of text with keyword embedded. (Keyword density too low?)
- 60 word block text with keyword at the start of paragraph. (keyword density even lower?)
- INDEX.html, a big page where keyword is mentioned once.
- Search for just Brandnewsacx1 returned the same exact result.
- A plain two word query returned "similar pages", where 8 results were flagged as similar. Why? Is it because Google is ranking by sentences? For example, 1-1 and 1-2 Both contain the keyword in a single sentence. Is it because...Google can't find related words for
Brandnewsacx1 sasahz? Why are those pages coming up as similar, when the description snippets are all unique? It could be just Google preventing a site from displaying more than 2 URLS/domain.
- Using site: the similar flag isn't coming up.
- In this result, Google is taking text snippets starting from the beginning of a sentence. So..Google is parsing documents by sentence. What the hell does it do with tabular information?
- Home page is now ranking above "unoptimized" page, so I assume that anchor text does have a boost effect, just below <B>.
SEO Test 1 Notes
- Maintain a certain density. For one occurance of a keyword, more words, lower ranking.
- When I use quotes around the query, the home page ranks higher. The rest of the SERP is identical. The home page ranks higher than unoptimized page. Why?
- Using site:seo4fun.com keyword returns the same SERP.
- the root places the lowest despite the fact that it's one directory above the rest of the urls.
- Some pages were 1k, but Google indexed them correctly anyway as unique pages.
- The page where the keyword only appears in the META keyword fails to rank at all.
- Page with keyword only in the meta description also doesn't rank.
- Create more pages that rank ABOVE unoptimized page. Right now, all I have is <title> and <b>
- meta description is not necessary for these handmade pages to get indexed as unique pages, because the onpage text is unique for every page.
- Order of SERP boost:
- No optimization I still don't get this. It's better not to use <h1>? Or am I abusing the tag and getting penalized for it? Does h1 also perform better if the keyword is in the following paragraph?
- <H2> This seems to prove H2 is weaker than H1 , since everything else is identical about these two pages. Or it can be penalizing the page more than H1(?).
- keyword embedded in a large body of text. This could be either due to keyword dilution or phrase "lost" in a body of text.
- keyword starts off a big paragraph. This could also be due to low keyword density, since this page is even bigger than the preceding page.
SEO Test Conclusion
Here are a few conclusions I can draw from what I've seen so far:
- META keyword has no effect if the keyword doesn't appear on the page.
- If a page contains a keyword only in META descriptions, the page won't show up in the MSN search results.
- For example, if I put "xxxfas98dhj" in the META description of this page, it will not show up for that search. I've taken a snippet from a unique description but it returned 0 results. A search for a unique text snippet that appears on the page returns results. This leads me to suspect that text in META descriptions for MSN is useless. Still, a keyword match in description will be boldened...so it may have an effect IF you also have the search text on the page.
- site:paysitestripsearch.com "review, promo galleries & pics - The first thing that'll hit" returns 0 results because the words inside quotes is not on any page text.
- site:paysitestripsearch.com review promo galleries pics The first thing that hit (without quotes/punctuation). Still 0 results.
- site:paysitestripsearch.com review galleries pics The first thing that hit (without quotes/punctuation). returns several results. Since I got rid of the word "promo", result of the words (review, galleries, pics) appear on the page. This is conclusive evidence that words in META description is useless unless those words also appear on the page text.
- I still want to know if keywords in META descriptions do anything to boost ranking in MSN. In the worst case scenario, META description is useless for improving MSN SERP
Other SEO4FUN Links
My Top Blog Posts