Link analysis is probably the one search engine optimization skill that most SEOs underdevelop. Their reliance upon link query operators at Google and Yahoo! ensures that SEOs have virtually no clue about what is happening with their linking profiles.
If you’re going to use a search engine as a reporting tool you cannot allow it to perform your analysis for you. If you’re going to use a search engine as an analytical tool then you cannot use it to perform your reporting for you. If you’re going to do serious link analysis then you have to learn to distill what the search engine tells you into useful information.
You have to use each search engine to analyze its own data. You cannot use Google to analyze Yahoo! and you cannot use Yahoo! to analyze Google. Until you learn that simple lesson you’re stuck in the “clueless and losing” category. You have to be confident in the data each search engine shows you. That means leaving “link:” and “linkdomain:” in the SEO blogs and forums where idiots and fools babble about them endlessly.
A good first lesson in link analysis is to use Ask’s search engine to tell you which page on your site it thinks is most relevant to “home”. Then ask Ask which page on your site is most relevant to your site’s name. Go through every one of your keywords and ask Ask which pages on your site are most relevant for those keywords.
How accurate is Ask’s assessment of your site’s structure? If it doesn’t come close to the mark (and I would expect no less than 90% accuracy from Ask for a well-optimized site), you most likely have a linking problem. If you have 100 pages on your site they cannot all be equally about peanut butter, not unless they are just absolute cookie-cutter spam pages.
When you have finished scoring your site for Ask, go to Live.com and perform the test all over again. Then go to Yahoo! and finally move on to Google.
Your results for all four search engines should be very similar. If you find huge discrepencies between their results that indicates your site is not being well crawled, which is usually a sign of weak internal linkage. So without ever looking at one link, much less one external link to your site, you can perform a test that tells you more about how the search engines see your site than any SEO tool on the market reveals.
That is the basis of truly useful, informative link analysis. But it’s only the beginning.
In fact, to do the first test correctly, you need a spreadsheet especially if you are ranking a site for more than a few keywords. I’m not talking about long-tail query feeding — I mean that if a Web site is being intentionally optimized to rank for more than 2 or 3 expressions you need to use a spreadsheet to study your internal linkage (at least until you get a feel for this style of analysis).
Once you have determined how strong your internal linkage is you should assess the value of your outbound links. There is only one valid test for determining if a link passes value in a search engine’s database: does its anchor text become associated with a destination page? If you outbound links are not passing link anchor text in any search engines, you have a problem.
PageRank hoarders disable as many of their outbound links as possible and in doing so they cloud their own abilities to test the power of their outbound linking pages. You should ask yourself: “If I have 40 outbound links on a page, how many of them pass anchor text?”
How well you can answer that question says a great deal about how much attention you pay to building useful links. Your outbound links are the only bellwhether you have for determining which of your pages is trusted, passing value, and helping other pages. If your pages don’t pass value to other people’s pages your pages cannot participate in the basic expert-hub-expert relationships that many people feel are vital to good search engine optimization.
Simply capturing links from so-called “hub” sites is not sufficient for topical relevance. You need to pass link value back to those hubs and/or to other experts. You need to associate your pages with the kind of content that you want people to associate with your pages. If you’re agonizing over “diluting your link juice” by linking out to a lot of sites, you’re not analyzing links, you’re playing with SEO myths.
Link analysis must tell you where your links are helping you (and other people) and where they don’t help anyone. You need to be able to analyze your own links before you can analyze someone else’s links. If you don’t understand the link structures you create for yourself both on and off your site you have no hope of understanding the structures other people make.
And there are no tools that will help you do proper link analysis.
If you have built your site properly, no search engine will ever know as much as you do about your linking structures. Search engine data will almost always be incomplete because they recrawl the Web (dropping old data from their live indexes) and because they don’t update their data for every site as often as the sites update.
So your knowledge of your site is the baseline against which you have to measure the search engines’ knowledge of your site. If they don’t include everything you need to focus more on getting the search engines to crawl your content and less on analyzing links on other people’s pages. The two do not go hand-in-hand.
Now stop and think about that. If your knowledge of your own site is the baseline you use to gauge how well a search engine knows your site, then what should be the baseline you use to gauge a search engine’s knowledge of another site? While it’s true that you cannot document every other site as well as your own you can learn to look for design features that are common to well-linked sites as well as design flaws that are common to poorly linked sites.
Use what you learn about the linking strengths and weaknesses of your own sites to develop a check list of do’s and don’ts that you can search for on other sites. That will help you with your competitive analysis and with your search for useful linking resources.
Learn to measure link anchor text. It will tell you more than you can possibly imagine if you have never done serious link analysis before. Real link analysis doesn’t crunch numbers. Real link analysis gathers data and organizes that data into patterns.
How you represent that data is your own business, of course. But the search engines are not link analysis tools. If you’ve just been looking at Toolbar PR, numbers of backlinks reported by Yahoo!, and where pages rank for specific keywords, you have yet to begin your linking analysis.
Read other articles in this series:
{ 5 comments… read them below or add one }
wibbler 09.24.07 at 2:44 pm
Hi There,
Ok – are you saying that the strengths of internal linking structures of a given site can be used to identify the linking structures which should be created between sites?
So if I have a page on my site which is doing really well for its phrase on my site, I should analyze the INTERNAL links to and from that page in order to establish / create similar patterns between pages on different domains?
Ok – scratching my head.
How the heckers do I get ask.com to tell me
“which page on your site it thinks is most relevant to “homeâ€. Then ask Ask which page on your site is most relevant to your site’s name.”
Is there a command to do that? Im not familiar with ask – but as a “student” wanting to become a master seo over the next 27 years before I retire, I want to understand EXACTLY what you are saying here.
It fits in nicely with what you said the other day about there being no link tool out there anymore, and it also fits in with a post you made about “rolling your own tools”.
Im a programmer – so I can roll my own – Im just short of a “spec” (when did I ever get a real “spec” anyway
Michael Martinez 09.24.07 at 4:31 pm
It’s a high-level point of view. It’s not about looking at links and finding a diagram you can apply to other sites.
Play with it. Try different queries. Ask yourself why a particular type of query shows you the results it does. Searching your own site is easier to learn from than searching the Web or other people’s sites because you know what your site is trying to do.
I’m not going to hand out any specifications, but if this week’s series of articles does their job, people will be able to see where they can improve their link analyses considerably.
Keep in mind that if you build a better link analysis tool and make it available for everyone else to use, you’ll get some links but you’ll ruin the tool for everyone.
wibbler 09.25.07 at 3:41 am
I never sold any of my programs – for that reason – and never will.
So no worries on that front.
Halfdeck 09.26.07 at 3:56 am
I’ve been looking at links for the past two weeks I feel like my eyes are gonna fall out. My focus has to do with neither anchor text nor PageRank, at least not for the moment. I look at two things: 1) the type of website linking out and 2) the reason a website is linking.
For example, take a look at this url: discovermainstreet.com/directory.php
The page, needless to say, is a directory. The website is a touristy site about Main Street, Sarasota. I’m looking at why this URL is linking to michaelsaunders.com. The reason is obvious: the Michael Saunders & Co. is located on Main Street.
How would I get myself on that page? Move to Florida?
Another example: truliablog.com/?p=147
Obviously, that’s a blog. Why is it linking to michaelsaunders.com? Because she just joined Trulia’s Board of Advisors. How do I get myself on board? Bribe Trulia?
Finally: zillowblog.com/real-estate-connect-day-1/2006/07/
Again, zillowblog is a real estate blog. And the reason for the link? Did someone pay Zillow $600 to write a rosy review of michaelsaunders.com? Nope. She was nominated for Inman’s Most Innovative Brokerage award. So I guess to get that link I’d have to bribe Inman?
There’s a big difference between those kinds of links and Text Link Ads in a blog’s sidebar. And the difference has nothing to do with anchor text, domain authority, domain age, or PageRank.
wibbler 09.26.07 at 2:08 pm
Its my belief that links used in se algos are a total farce these days.
Given that the SE’s define what is “natural” – by definition – it is them “creating natural” which makes it totally unnatural – and therefore a farce.
In my book – the engines have made it “natural” to sometimes HAVE TO purchase links – therefore any attempt to thwart purchased links is unnatural.
The days when truly natural linking are gone for now.
Do whatever it takes – which may or may not include purchasing or bribing for links.
One things for certain – the knob twitchers will twitch their knobs again soon enough no matter what happens….
You must log in to post a comment.