Advanced On-Page Optimization - Whiteboard Friday

Posted by Kenny Martin

In this week's Whiteboard Friday, Rand goes into depth on how you can optimize your on-page content. Presented here are five advanced tactics that get you thinking beyond the basics of traditional page optimization and set you up to start creating content that's both relevant and unique.

How do the phrases on your page relate to one another? Does your page content make your visitors happy? By moving beyond simply making sure those title tags are optimized and scattering a few keywords around, we can produce intelligent content that's not only engaging but gets better rankings too.

As promised, here is a Quora thread that describes Google's use of Machine learning.

Video Transcription

Howdy, SEOmoz fans. Welcome to another edition of Whiteboard Friday. This week, we're talking about advanced on-page optimization. Specifically, I have five tactics for you that go beyond the traditional "I'm going to put my keyword in the title tag. I'm going to put my keyword in the URL", those kinds of things.

First one, starting out is this idea of semantic connectivity. We talked about this in the past. We did some research a couple of years ago, maybe 18 months ago on LDA, which is latent Dirichlet allocation, which, of course, is a form of topic modeling that we suspected Google might be using.

It's a way to imagine the connections between words in a particular language. I'll give you an example. Here is the word "cat", and the word "cat" is probably closely related to the word "feline". If you were a search engine and you saw a document with the word "cat" and the word "feline," you would think that document is more relevant to a query for the word "cat" than a document that has the word "cat" and the word "whiteboard," which maybe that would be very far away.

Here's cat and here's canine. Those are much more distant, but cat is relatively closer to tiger, but it's even a little closer to meow. So, you get this sense of, ah, the search engines have a graph of all the words in a language set, how they're connected to each other, what's relevant to what, phrases not just individual words but the two or three or four word phrases.

This kind of thing can be very helpful if you're looking at a document and you're saying to yourself, "Boy, I talked about cats, but I forgot to mention anything about what they eat or what family they're in or what they're related to. I didn't even use the word 'pets.' Maybe, I should be optimizing for those types of things." Employing those closely connected terms can help to boost the relevancy and help boost your rankings.

Second thing on the list, block level optimization. There is a great YOUmoz post about this that we promoted to the main blog recently talking about precisely this type of thing where search engines will essentially analyze individual portions of a page. They'll look at, oh, here's a sidebar and we've decided that's not really relevant because that's navigational links or here's the top nav. We're not going to analyze that for relevancy as much potentially. We're going to look at the header of the document, where the headline is, those first few sentences. We're going to look at the middle of the document, maybe in paragraph forms, the footer of the document, the end. Are all of those things talking about the topic? Are they all on the subject, or is this something that starts out talking about tigers, but it eventually gets into a discussion on genetically modified foods? If that's the case, maybe it's less relevant to tigers. It's just that the initial headline looked like it was relevant to tigers, and so therefore, we don't want to rank this document for the word, tigers. We might even want to be ranking it for something like genetically modi