April 2003

April 2003

Year SEVEN
Issue Four

Planet Ocean Communications

Forget about linking all your pages together equally…
Precision Guided Linking is Key to Top Results on Google
by Leslie RohdeGetting top position at Google depends almost entirely on two factors – link reputation and PageRank (PR). The impact of link reputation has been discussed a number of times within this publication (see: Feb’03 & Sept’02 articles). Rather than cover old ground, this article will focus on PR and how it can be controlled through advanced, and very precise, linking techniques.

First, one should bear in mind that, used incorrectly, these advanced linking techniques can create a positioning disaster – hence the standard advice to avoid their use. But, when used with an understanding of how Google actually works, the advanced linking approach shown here provides an incredible ranking advantage.

Three Kinds of Links

Functionally, a link is a way to get from one page to another. The most basic type of link is one we are all familiar with and looks something like this:

<a href=”path to page”>link text</a>

We’ll call this a text link. As we’ll see, this is the most important form of linking for search engine purposes, but by no means is it the only choice available. There are two other types of links that we may employ.

The next type, and closely related to the text link, is the image link. This is composed of an HTML <a href=> tag just like the text link, but instead of link text an <img> tag is used like so:

<a href=”path to page”><img src=”picture file” alt=”some text”></a>

This type of link is often used instead of a text link because it can be made more visually appealing to human visitors. The problem, with regard to search engine rankings, is that image links are almost always a poor choice. That’s because their lack-of-text does not give the search engine any idea what the link is actually about. Simply put, search engines cannot interpret the meaning of images.

The final sort of link is a striking departure from the other two and is a relative newcomer to the web. Instead of using the <a href=> tag in HTML, we use Javascript functions to redirect the user’s browser. Javascript navigation can be made to look and act just like text and image links to humans, but the search engines cannot follow this type of link. This can be a problem or an opportunity depending entirely on how this technique is employed.

A simple Javascript link will look something like this:

<img src=”picture file” onclick=”window.location(‘path to page’)”>

We’ll call this a dynamic link to distinguish it from static HTML links.

The Four Purposes of Links

Before we can decide what types of links to use to connect our pages together, we need to understand what links do. Links provide four different functions for web site design – and only one of them relates to human visitor experience. The other three concern our placement in the search engines. In summary form, here are the 4 functions of links:

  1. provide navigation for humans
  2. provide spiderability of site for search engines
  3. create link reputation
  4. create link popularity and distribute PageRank (PR)

Let’s consider each of these functions in turn.

Human Navigation

The purpose of links for human visitors is pretty obvious – the links provide the means by which a visitor can move around our site to find the information they are interested in. Often, the visitor may reach our site using one set of searches and find something related that he is actually more interested in. Links provide the means to draw a visitor deeper into our site. Links also provide the pathways to lead the user through the sales process. It is by no means an exaggeration to say that links are an indispensable part of e-commerce.

Spiderability

Our pages have to be spiderable. What this generally means is that every page of the site must be only a small number of “link hops” from the site’s index page if we expect the SE spider to find them all. That’s because it’s the links within the site that the spider uses to find our pages in the process of adding them to their index. If we don’t provide links for the search engine spiders to follow, our pages will not get discovered nor will they be indexed.

Link Reputation

The text used in links is utilized to compute the reputation of the page. Similar to the analysis of page content (known as the “topic” of the page) reputation is a textual analysis of what other pages say the target page is about – via the links used to refer to the page.

The text that Google (and the other engines) analyze to create the reputation for a page is the text placed between the <a href=> and </a> tags on linking pages. The diagram below shows the page in the center with 3 incoming links to it. Although the page’s text is not shown, it would appear, based on the content of those incoming links, that the page is mostly about widgets.

The page in the center appears to be about 'widgets' based on the text-content of the majority of the links pointing to it

This is precisely the sort of reputation analysis that search engines do when establishing the ranking of pages within search results for keyword searches. And, you can do your own reputation analysis by using OptiLink

Link Popularity and Page Rank

But link reputation is not the only measure used in positioning pages. All modern search engines use the number of links to a page as a measure of the page’s overall importance on the web. At Google this measure is computed by their proprietary PR algorithm, while other search engines use a far simpler formula.

PR is a factor of all incoming links combined

PR is a recursive computer algorithm that computes a numeric value for every page in the Google index based on the PR of other pages and the linking relationships between those pages. Simplistically, the PR of a page is the sum of PR contributions from the other pages that link to it. Consider the diagram on the right ==>

Of course, these are not real PR units and the damping factor isn’t considered (Real geeks are invited to read the original paper by Google’s founders), but for positioning purposes, we really need only understand PR conceptually. In other words, the precise numbers are more relative than they are important whenever we are presenting the concept in an overview.

What’s important to understand from the diagram is that the contribution a page makes to each of the “linked-to” pages is an indiscriminate distribution. In other words, from Google’s perspective, the page cannot play favorites and the more pages it links to, the less PR each linked-to page receives.

The more outgoing links on a page the less value those links carry in computing the PR of the pages they point to.

For example, let’s take the previous diagram and remove one of the outbound links. Now instead of the PR=20 being spread among 5 pages, there are only 4, resulting in a higher PR factor being applied to each of those pages. By understanding these two diagrams you’ll understand why link farms are so ineffective – too many links!

Different Links for Different Purposes

The table below summarizes the three types of links and shows which of the four different features of links each provides.

Link type User nav Spider nav PageRank Reputation
dynamicYesNoNoNo
imageYesYesYesNo
textYesYesYesYes

As stated earlier, top positioning at Google depends on reputation and PR. Therefore we can deduce there are certain combinations of linking that will almost never be productive, ranking-wise, for pages that we expect to get positioned well in the search results. Image links are a prime example.

On the other hand, there may indeed be some pages on our site that we don’t need to have positioned in the search engines. For such pages we can certainly consider using links that don’t “contribute” to PR and reputation. In fact, that’s precisely the strategy we will illustrate.

Three Kinds of Pages

Nearly every web site, except the smallest of micro-sites, will have three sorts of pages:

  1. pages we would like found for our primary search term
  2. pages we are positioning for secondary search terms and
  3. supporting pages that aren’t dependent on SE positioning but are necessary for the overall operation of the site.

The home page of our site is almost always the first sort of page. It typically relates to the broad topic of our site and we’d like to have it well-positioned in the most competitive search relevant to our topic. To competitively succeed, this page will need the highest possible PR and the best link reputation that we can get.

Generally, there will be a second set of pages we’ll want positioned for lesser search terms. For example if I’m selling widgets that is the likely subject of my home page (and probably a more competitive category) while blue widgets or custom widgets might be the topic of my internal pages and the keywords that we would focus on positioning well for their respective searches.

A web site is often, and correctly, designed based upon a hierarchy of presumed natural relationships that can be decomposed (broken down) into sub-components.

An example might be that auto repair could be decomposed to engine repair, body work, suspension, interior, and so on. The top level subject will almost always be a more popular search than any of the sub-topics so we’ll want to use a linking strategy that makes our home page the most important page on our site.

The third category of pages is not intended to be positioned in the search engines at all. For example, a “privacy policy” page is one intended for those who need that sort of reassurance after they have found the product or service they were searching for. Nevertheless, it typically isn’t a page we’d care about ranking well in the search engines. Other examples would be “about us” or “contact us” pages although they DO provide us an opportunity for linking back to our important pages. For example, a privacy page could easily be used to link our home page using a keyword-rich text link.

Typically, the PR of a home page will increase as the number of internal pages increase assuming they all link back to the home page. Regardless, “optimization” is about getting ranked better and faster – so let’s take a look at how the correct linking strategy can speed up the process thereby giving us a real edge!

“Standard” Linking

A typical home page might look like this

Let’s start with a “standard” website. We have a home page and 3 sub-pages related to different parts of our topic. All the pages are generated from a template with a navigation bar along the left side. We’ll assume that text links are used for these navigation links. The illustration to the right depicts what such a home page might look like.

The internal pages (3 of them) will have the same 4 links along the left edge of the page. In network jargon this is a “fully interconnected” site, because each page links to every other page of the site. The diagrams (right & below) illustrate this. Without considering external links to and from our site, the number of links to and from each of our pages is the same, so the PR of all our pages will likewise be the same.

A typical home page might look like this

But, is this what we want? Generally not! As noted above, the sub-pages will typically deal with a more narrow topic than the home page and will therefore not need as high a PR value nor as much link reputation. It is our home page that needs the most help, but the linking strategy suggested to the left has placed it on a par with our internal pages. If we could divert some of the PR from our internal pages to our home page this would improve our chances of winning top placement for the home page and probably without sacrificing the positioning of our internal pages. By making careful use of dynamic links we can do just that.

Linking Our Way to Higher PR

First, consider what would happen if we made all the links in our site dynamic links instead of text links. That would be a disaster. The lack of any spider-sensible links would render the internal pages of our site inaccessible from the home page by a search engine spider. That means that Google would only find the home page and it would have a very low PR as a result.

But, that’s a problem we can fix with a site map by providing an “ugly side door” (i.e., text-link) for the search engines while leaving our flashy front door open (i.e., dynamic links or image links) for human visitors. Therefore we should always build a site map linked from the home page and linking back to the home page. This gives the spider a two hop path to every page on our site (via the site map) and leaves us free to mix and match any linking styles we want without worrying about spiderability.

Furthermore, since the site map MUST use <a href=> style text-links to each page, these should be text links containing the keywords most relevant to the target page. One simple approach is to use the target page’s title text, since it too should be a keyword rich description as well.

Now, even though our site map provides good spiderability, we’re still not done. Even with all our pages indexed, they still don’t link to the home page as far as the spider is concerned because the spider can’t see dynamic links. Not cool. But suppose we use a text link to reference the home page and dynamic links to reference the internal pages? Voil�! …the home page gains the link reputation and enhanced PR as a result.

Here we’ve added the site map so all pages get spidered and changed our navigation bar so that a text link is used for the home page and dynamic links for all the other pages. The diagram below shows the spider visible linking structure for our site. Humans will continue to see and use the graphical navigation bar (implemented using Javascript) while Google enters our site via the “servant’s entrance” 🙂

Google’s view of our linking structure —
notice the one-way links.
This is the link structure seen by the search engines
Human view of our linking structure — notice the two-way links.
This is the link structure seen by humans

We now have an linking structure designed esthetically for humans that is essentially invisible to Google! The site visitor sees a highly interconnected site while Google’s spider (googlebot) sees a far simpler structure. The effect this linking structure has on our home page’s PR is significantly beneficial. Our home page should have a PR some 6 times as large as the PR of our internal pages (stated in linear units, not “Google Toolbar” units, but again, the precise values are relative rather than specific).

Precision Guided Linking

Using the right kind of linking, we can provide a rich user experience while also controlling reputation by guiding the engines toward those pages we want them to consider as “most important” without resorting to the “equal page linking” approach that has been favored in the past.

And, just as we don’t want our internal pages to be as important as our home page, we probably don’t want all our internal pages to be of equal importance either. By using a mixture of text links and dynamic links, intermingled in the same navigation bar, we can emphasize some of our more important pages while de-emphasizing our lesser important pages (think: privacy and contact us pages).

The rules are simple…

  1. Provide site wide spiderability with a site map. This adds to the reputation of all of our pages while providing easy spider access so we don’t have to make our human navigation features spider friendly.
  2. Use only text links for pages that you want to get positioned well in the search engines and always use the keyword phrase that is most applicable to the topic of that page as the text of the link itself.
  3. Use dynamic linking to provide human navigation for supporting pages that do not need to be positioned well in the search engines so that you do not adversely distribute importance (i.e., PR).

Conclusion

The two most powerful SE positioning forces at Google today are Link Reputation and PageRank. This is likely to remain constant for the foreseeable future. The optimization of both of these measures is best accomplished with linking techniques. On-page elements are still considered, but they are of lesser importance.

By using advanced HTML features and the scripting capability of modern browsers, we can now provide a rich user experience while at the same time offering an optimal linking structure to the search engine spiders. The advantage to our business is that we’re able to accomplish more with fewer pages. This multiplies our positioning power and stretches the reach of our content to garner a greater return on investment with less effort.

Best Regards,
Leslie Rohde
Leslie Rohde – Windrose Software
Creator of OptiLink


� Copyright 1997-2003 Planet Ocean Communications, Inc.
Planet Ocean is a registered trademark of Planet Ocean Communications, Inc.

Biographical and Press Packet for Leslie Rohde