There are many definitions for “cloud” in the information management and telecommunications industries. On the Web most people think of tag clouds when discussing technology clouds because tagging clouds have become commonplace on blogs and shared resource sites.
Tags are used to identify keywords or internally indexed words. Your blog may create a tag cloud based on keywords you associate with each post. I don’t create tag clouds for SEO Theory but if I did the Wordpress software we use would figure out which keywords carried the most weight by counting the number of posts that are tagged with them.
In technology, a cloud is any data conduit that accumulates random data structures from multiple sources; stores, processes, or transmits them by similar means; and then dispenses them in their original state. The metaphor is obviously based on real clouds, which are insubstantial physical structures made up an large numbers of smaller substantial structures.
In search engine optimization we can speak of, and manipulate or manage, clouds of many different types. Tag clouds are the only common type of cloud that the SEO community discusses, but there are other types of clouds that we have used for years.
For example, have you ever put together a multi-site search tool? I have, more than once, and I know of many SEO search tools that have been built from Google Custom Search and other virtual search engine components. When you collate URLs for a customized search function you create a web site search cloud.
A fair number of SEOs today are exercising their reputation management skills by filling out social media profiles for themselves. Now if you search on these SEOs’ names (or screennames) you’re likely to find a lot of little profile pages. These profile clouds don’t really serve any useful purpose and are a low-level form of search spam that obscures your ability to see relevant information about people. We could say these profile clouds are the effects of creating an SEO fog campaign.
While not very elegant or challenging, fog campaigns have been used for both hostile and defensive optimization. The only effective way to disrupt a fog campaign is to drive the fog sites down with “hot” sites that a lot of people link to. Call that a hot campaign, or a wildfire strategy, where you set some Web sites ablaze and watch the heat spread from site to site. Of course, good link baiting helps wildfire SEO considerably.
There are also link clouds. I have seen two types of link clouds so far, but don’t underestimate the inegnuity of the link spam community. There may already be more than two types, or perhaps there soon will be.
A Type I Link Cloud is a set of links based on a single domain, but these are not site-wide links. Think of a site like CNN that hosts many thousands of articles. Among those articles may be a fair number of links to some specific sites, such as the SPCA or the White House Web site. To be considered a link cloud, these non-promotional links must be distributed randomly across relevant content. They accumulate gradually and don’t appear all at once.
A Type II Link Cloud is a set of links based on many domains, but these are not brand links. That is, any link pointing to SEO Theory with the anchor text “SEO Theory” (or “Michael Martinez”) would not be part of a link cloud. The cloud aspect is derived in part from the random distribution of the links and from the randomized text that is relevant to what we could call a core cloud topic. That is, if you took all the links that comprise a Type II Link Cloud, their anchor text would be similar enough that you would know they were all pointing to the same thing.
A Type II Link Cloud may share link nodes with other Type II Link Clouds, but the majority of the links are unique to that cloud. For example, there may be 100 links pointing to SEO Theory’s “20 Hard Core SEO Tips” article from last year. Any of those 100 links that use my name or the blog’s name as anchor text are ignored. The remaining links constitute a link cloud that points to the article.
The less variation there is in your link anchor text, the less like a cloud your links become, so randomness is a necessary aspect to the link anchor texts. If you were to visually represent a link cloud using different font sizes, it would look exactly like a tag cloud. A natural link cloud includes multiple locations and anchor texts, all pointing to the same topic.
Search engine optimization can look at link clouds to gauge how natural a link profile may be (although we have no knowledge of what boundaries the search engines set for natural link profiles in their own algorithms). You can get too caught up in the link cloud paradigm, but visualizing your backlinks as a link cloud may help you see just how artificial your linking efforts really are.
There are other types of clouds, too. One can easily speak of content clouds, which cannot exist on siloed Web sites. A Web site that follows the siloing paradigm arranges all of its data by topic and doesn’t allow for duplication of branch topics. A cloudy site may arrange its data by topic but will nonetheless allow duplicate topic branches. So today you may create a site that talks about dogs, cats, horses, and bees. Tomorrow you may add a second dog section to the site, rather than simply add more content to the existing dog section.
A cloudy Web site has some advantages over a siloed site, but the siloed site does at least provide both the Webmaster and the visitor a clear hierarchical tree to walk. If you build or optimize cloudy sites, you’ll need to make sure your on-site optimization includes site search and/or tag clouds. However, blogs are not intentionally cloudy, especially not tagging blogs, since the tags usually work with tag pages that index or duplicate the tagged content.
You can have page clouds or section clouds but you cannot include pages in section clouds or sections in page clouds. That is, page clouds and section clouds are mutually exclusive data sets. You can, however, use either page clouds or section clouds to demarcate your topic branches (because they are mutually exclusive).
In fact, any two or more mutually exclusive cloud formats may be used to map the same core data. This is important to remember because your cloud definition sets the context for the data it includes. Data that does not match the context is not included in the cloud. All the SEO Theory pages that include the word “horse” do not appear in the cloud that consists of all the SEO Theory pages that talk about something other than “horse”.
And I should point out that you cannot define a cloud on the basis of what’s NOT in it. That is, the description of the cloud’s content must specify something that IS in the cloud, even if the something is not associated with something else. There is no null cloud, no empty cloud. To be a cloud, it must contain two or more nodes that match the criteria for being included in the clouds.
These clouds are, of course, just sets, and I’ve discussed set theory before. A set is any collection of objects together with precise criteria for determining whether or not a given object is in the collection.
You can shape the clouds (or sets) to work toward specific goals. But that’s an article for another time.
{ 0 comments… add one now }
You must log in to post a comment.