(photo credit: http://www.flickr.com/photos/eirikref/)
This post on Marketing Insight is how to build your site. (And those are silo's above - where grain is stored before it is shipped to processors.) Dr. Williams (below) is offering one concept about how to build a site. He's actually converting old hand-coded sites over to a new WordPress platform. And I respect his work.
ezSEO Blog: "The overall idea of this new system is to concentrate on sub-niches, one at a time, and build those sub-sections before moving on to the next one.I would take a different approach. And as you know from following this blog I use RankTracker to query WordTracker to get my niches (which is a whole lot cheaper and more powerful).
THINK OF EACH SUB-SECTION AS A MINI-SITE
Suppose my website was on dieting. This is a huge niche and to try to do the keyword research up front would be a monumental task. A large part of my site would probably be the various diets that people could go on, so I would start off by setting up my Wordpress site with a super category called 'Diets'.
I would then pick one diet at a time, e.g. the Sonoma diet, and carry out the keyword research only on the Sonoma diet. I’d keep this data in a separate database in KRA Pro.
I could then concentrate on the Sonoma diet, creating a main page for that diet as well as articles on the diet that can link back to the main page.
When you are finished with the Sonoma diet, pick another area you want to work on, e.g. South Beach Diet and repeat.
IT’S LIKE ADDING LOTS OF MINI-SITES TO THE SAME DOMAIN, AND LINKING THEM WITH THE HOMEPAGE AND MENU SYSTEM"
Now, as I've discussed, my approach is to distill my niches and their keywords. This gives me what keywords add up to the targeted main keywords I want to use to create content. I simply line them up by KEI and then work on them in that way. I don't take a certain one and then create content for all the pages like that. I don't do all the dog-collar keywords and then do dog-leash keywords. I'd work all of the long-tail niche keywords for dog in order of their KEI. Sure they'll cross, but what you are trying to take over is "dog" as a main KW - and that is what your blog is named. (Yes, that is a lousy choice as a "niche", since it isn't really, but it gives a good example. )
Under dogs, you'd have categories (silos) of dog-collars, dog-leashes, dog-dishes, and so on. As Andy says above, I don't really hold to silos either. Mainly because people don't want their content served up that way. And your "back" button is there by default on every page you visit. So bouncing from a too-content-limited page is easier than not.
And WordPress has the option of viewing all your categories on the sidebar. So while they might be interested in dog-collars, they might want to compare with the content you have for cat-collars. Or just Cats. Those are all on your categories - and allow you to nest categories as well.
I recently imported a Blogger blog into WordPress and found that now I had tons of default categories. So that screwed any idea of having category-silos without editing every single page out of hundreds. (And I have more to study up on the use of both categories and tags for posts - both of which cross-link posts, making it easier for viewers, but ridiculous if you are trying to maximize page-rank.)
Another reason I like to post by KEI rather than category is to break up my week. While I still have lots of research to do on the various keywords, it gives you more diversity and options if you are posting for the best traffic/competition first. And when you have all the long-tail keywords established, you can come back to work them all in sequence again (or several times) because you already have the research done.
My writing is what I am inspired on - so I jump from blog to blog, depending on what I'm covering at the time. Lately I've been hobby-horsing how to stop telemarketers' Internet scams, but this also gave me a post on hate-addiction, as well as putting up a new post from an old draft about expanding your marketing mix beyond email newsletters I had hanging around on that imported Blogger site. So I'm working on several keywords at the same time. Each with different publics. Keeps me from getting bored, but it's mainly to get those-type thoughts out and written down before I forgot about them and lost the inspiration. (Plus, it makes my friendfeed life-stream far more interesting, let alone twitter.)
And this method of writing for just a single blog also makes more sense to the search engines, since a person doesn't just talk about dog-collars for twenty articles over a couple of weeks and then fascinate on dog-leashes. This also breaks up your flow if you are doing posts, so your subscribing public isn't bored to tears by dog-collars, then bored by dog-leashes. A simple approach would be to write in one category, then do your next post in the next-best KEI category, and so on.
But I'm pretty sure with both tags and categories, any idea of a silo is pretty shot. Silo's are actually a hold-over from the days when PageRank ruled SEO. And pagerank has been pretty back-watered for some time now.
The general rule is that search engines follow viewers. So Content is King - like nothing else. Build a great site and you'll get more subscribers and they'll stick around for more (as well as buy).
The point of doing your articles/posts by KEI then targets your most likely traffic first. Rotating through these keywords until you have several articles for each long-tail-KW - and a good leg-up on taking over the main phrase, this keeps your readers interested and coming back for more.
Just my thoughts on the matter. What are yours?