A friend of mine is creating a how to website. I recently wanted to purchase an item related to his field, so the first thing I did was go off to his site to look for information about the product I wanted to buy. His article about the topic was quite thin in nature and perhaps even looked like I might have wrote it, which is bad since I know nothing about the topic.

If someone is buying a cheap accessory then you might not have to sell much or sell hard or go in much depth to convert them. If someone is buying something

• that they know little about

• is expensive (in terms of opportunity cost - time, money, other factors, etc.)
• is hard to return

then you might have to provide more information to be able to build up enough trust to sell to them.

When creating an affiliate database driven site it is easy to give 1,000's of items the exact same weight, but if you can instead answer one or a few questions far better than anyone else does it is much easier to create a longterm stable income stream. Plus if most competing sites consist primarily of thin compacted data and your sales information and product guides are link worthy that provides a huge marketing benefit.

Also consider that as search engines dip further into vertical search and more thin compacted data sites are created one needs to provide better or more unique and compelling information to be citation worthy.

AOL has a bias toward consumer (ie: non b2b) type queries, and they may have a higher % of brand related searches that both act to place a bit more emphasis on the top search result than general searchers from other search engines, but some people have dug through the 20 million search queries AOL gave away and come up some stats.

SEO Blackhat dug through the numbers and has a free tool for estimating clicks based on search volume estimates and rank position.

I only attended a couple panels at SES, but Greg Boser was on one of them, and he always has a way of saying things in a clear way. He mentioned in the past that a divide and conquer technique was a great way for small sites to compete with larger rivals. He then went on to say that with Google's current reliance on site age and link related authority that it may no longer make sense to use a divide and conquer method to rank well in Google. If you look through Google's search results for competitive insurance related phrases typically they are dominated by old sites, government and education sites, news sites, and/or sites which are focused on all 50 states. In the past it might have made sense to make sites for each of the most important states, but with the current Google one site with an authority rank of 8 is probably going to be worth far more than a half dozen sites in the same vertical that only have an authority rank of 6 (there is no AuthorityRank meter...just assume it is some arbitrary value based on age and link equity).

Another thing which Greg mentioned in his speech was that it seems Google is really moving away from trusting anchor text as much as they used to. I recently bought an old domain from a friend that was just wasting away. It was old and had a few average type links from related websites, but had no relevant anchor text for the terms I wanted to rank.

I changed the internal link structure to focus the home page on a moderately competitive term. Just doing that ranked it in the top 20 for that term. I then got it a couple low-to-average-quality links with the plural version of that anchor text and got it ranked in the top 10 for both versions. In the past that site might have required either higher quality links or many more descriptive link anchors to rank.

In the past (say a year or two ago) I was way more focused on getting specific anchor text from external sources and probably went a bit far with it. Now it seems all you need are a few relevant decent quality descriptive links and your site will rank so long as your site has a bit of age and a few legitimate links.

Large websites tend to have many useless pages associated with them. They may be caused by any of the following

• poorly structured or poorly formatted user generated content
• content duplication due to content management issues
• canonical related issues
• dangling nodes which act as PageRank sinks
• navigational pages which are heavily duplicated and soak up link authority and do not provide a clean site structure

I recently have had a couple SEOs show me various navigational techniques which made thousands of thousands of somewhat similar mid level navigational pages.

Some pages make sense to be indexed and provide a great user experience if searchers land on them. Others provide a poor user experience.

Search engines do not like indexing search results from other engines, so if your navigational scheme has an element which acts similar to an internal search engine you probably do not want all those search pages getting indexed if they are heavily duplicates of one another.

I was talking to Stuntdubl the other day, and he stated one of the main things he likes to look at to get a general indication of the health of a site is to look at the ratio of quality pages indexed to total pages indexed from your site.

If lots of your indexed pages are heavily duplicated and/or of low value that may cause search engines to crawl or index your site less deeply and not index all your individual product level pages.

A few people have created free cool web based tools which allow you to search through the 20 million keywords AOL recently shared with the marketing community. http://www.aolsearchdatabase.com/ - allows you to sort data by:

• User ID
• Keywords
• Date of search
• URL

http://www.askthebrain.com/aol/

• allows you to sort data by TLD
• allows you to sort data by keywords leading to a specific domain
• by default the tool also displays the top few thousand URLs, the number of referrals to a URL, shows the top keywords leading to a URL, shows the keyword diversity ratio

http://www.dontdelete.com/ - search by keyword, keyword stem, or part of a keyword to find related keywords. For example, if you searched for dati it would return all keywords that had dating in them.

Anyone think I should add a link to any of these tools on my keyword research tool?