When I was new to SEO I did a bunch of on the page analysis to try to figure out exactly what other people are doing. The problem is that it gets you focused on things that do not matter. A site may end up ranking high at the sacrifice of conversion.
As search algorithms advance basic link analysis tools, at least for Google, are starting to become what keyword density tools are: a waste of time.
Link analysis software was cool, especially when Google used to show all of the PageRank 4 and above links, back when their search relevancy algorithm was a bit more dependant on raw PageRank.
Now Google only shows a limited random set of backlinks, and the other search engines also limit the search depth to 1,000 results, which makes it hard to do useful analysis with the various link analysis tools on the market.
If it were quick and easy to query a database deeply (deeper than 1,000) then the link analysis tools would be much more useful. None of them currently on the market really make that a quick and easy process.
The other problem mentioned above, of harder relevancy algorithms, is not one that is going to go away. WMW supporters forum has a thread about the Google "Sandbox". A friend of mine just got through reading the whole thing, stating that most of it was of little value, but he found a couple posts enlightening. The second post here by Ron Burk has some good tidbits
To keep improving the results, you find more variables for the algorithm-creating machine to use, and you add to your store of human-ranked pages for it to "learn" from. What you don't do is bother understanding the actual algorithm -- it was constructed by a machine and is way too complex for anyone to keep in their head.
...
Psychologists have shown repeatedly that when you give people a system to optimize, all you have to do is secretly introduce a delay between their actions and the results of their actions, and they will go bonkers. In fact, in a very simple (single variable!) model in which people are trying to control the temperature in a virtual refridgerator, you can get some of the same irrational responses you see in these forums
and the first post here by Captain CaveMan (which incidentally is the name of an awesome cartoon character) does as well:
Without giving away the store, I don't know how else to say it. There is no sandbox. People speak of it as though it were some simple 'thing' that stops new sites from being seen. That has simply never been true. What was true was that in its early days, some of the algo elements and related filters were so tight that only a very few new sites got past them (some accidentally; some methodically). Over time that changed; more sites started getting out, presumably as G worked to surface more new, higher quality sites.
There is no sandbox. There is only a serious of rotating algo's and related filters, that make it far harder for sites launched after spring of '04 to be widely seen in the SERP's. Not impossible. Harder. And certainly not as hard now as was true seven months ago. This has been hashed and rehashed so many times that it's hard to understand why it's still confusing.
If you can only see a few of the variables and overexert effort to satisfy those variables you may end up tripping filters and not satisfying other criteria.
