Link analysis done right

SEO bloggers like to talk about links and promote their friends’ link tools. They also like to provide you with long check lists of things to examine for your link analysis.

Most of what you’ll read on the Web — particularly from SEO bloggers — concerning link analysis is time-wasting nonsense.

Every search engine treats links differently. And that is why you cannot use any one search engine to perform link analysis for any other search engine.

Every SEO link analysis tool looks at Yahoo! backlinks and Google Toolbar PR, and that is why you cannot use any SEO link analysis tools. They aren’t telling you anything useful.

If you’re going to use snapshot SEO analysis then you need to set limits for the data you’re going to look at and you need to set boundaries for the conclusions you’re going to draw.

Yahoo! won’t ever be in a position to tell you how powerful a site is in Google. Every idiot SEO “A list” blogger in the world may be telling you otherwise, but they’re either just blowing smoke out their ears or they don’t know what they’re talking about. Yahoo! backlink reports do NOT tell you what Google knows about your link profile.

So, if you want to know whether a Web site MAY have a lot of links in Google, look at Yahoo!. If you want to know whether a Web site DOES have a lot of links in Google, spend the next 2 weeks sifting through Google search results. There is no other way to determine what a Google backlink profile looks like.

Now, you can use Yahoo! and any other search engine in tandem to see if your Yahoo! links show up in other search engines, but you have to be careful NOT to assume that the links are passing value. Each search engine puts restrictions on links and they don’t use the same rules. Remember that as many as 60-70% of all links do NOT pass value in Google. Remember that Yahoo! claims it pays more attention to the “first” link from a domain than to the rest (has anyone ever explained what the “first” link criteria are?).

Google Toolbar PR takes it on the chin every time I talk about link analysis. To this day I have not found a statistically valid reason to look at Toolbar PR data in analyzing backlink profiles. Just because a page was given a Toolbar PR value of 6 three months ago doesn’t mean I should be impressed today. Nonetheless, if you TRACK Google Toolbar PR value over time, and if you have data covering a year or more, you may be able to identify significant trends in linking page value. You want to see relatively stable or gradualy improving Google Toolbar PR values if you’re going to look at them. Don’t agonize over whether they are above or below certain numbers.

What you want to know about linking pages (aka link sources, link locations, and link inventory) is:

1. How often each page is cached in each search engine
2. Whether the links on the page pass indexing value
3. Whether the page operator is responsive
4. Whether the page is isolated or otherwise statistically likely to trip some sort of filter

Link theory calls for a value-based hierarchy in links. That is, the more types of value (eyeballs, clicks, anchor text, pagerank, trust, crawling) a link passes, the more important the linking page is to a Web marketing strategy. The fewer types of value a link passes, the less important the linking page is.

Frequency of cache date by itself only indicates that the page is refreshed in the index, not that the page will pass value through its links. Passing of link anchor text only indicates that you may get some anchor text but not when. The combination of cache-frequency and anchor-passing gives you a better indication of linking value than all the SEO tools in the world.

You cannot determine caching frequency with fewer than 3 cache dates. You don’t need to capture every change to the cache, but you do need to establish a rate of cache-change for each linking page. You decide for yourself what the acceptable minimum rate of cache-change should be. Pages that change every hour or day may not be optimal for long-term linking strategies. Do you have an idea of why that may be so?

Relevance is not very important if you’re just looking for crawling, PageRank, or anchor text but it sure helps if you want to maximize the value of your link building efforts. The more relevant links you have from high traffic Web sites the more likely you are to get referrals from those sites and the less dependent upon search traffic you become.

Your goal as a search optimizer — where link building is concerned — should always be to minimize a page’s dependence upon search for traffic. Although that seems counter-intuitive, if you only build links for search you’re not doing yourself (or you client) any favors. We’re optimizing for search as a part of a larger plan, to draw qualified, interested, converting traffic from as many sources as possible.

Efficient link analysis takes time. The slow, measured pace gets you results faster than building links in volume or checking every page you look at with an SEO browser tool because you’ll eventually get to the point where — as you carefully evaluate pages one-by-one — you can intuitively deduce whether a page is a useful link source just by looking at it. The feel of a powerful, value-passing page is very different from the feel of a weak, spammy page.

One of my history professors in college said you could always tell which kings and rulers were the most powerful men of their generations and which ones were flamboyant wannabes who had to placate all their neighbors.

Truly powerful leaders usually opened their correspondence with their names: “I, James, King of England, send greetings and felicitations to my friend Prince Willhelm, ruler of the low territories, defender of the weak.”

The weak leaders always wrote more about themselves in their openings — often more than comprised the rest of their letters: “I, Lord Prince Edward, Duke of Hemmingway, Earl of Sebastien, High Ruler of the Five Isles of Servitude, Keeper of the Sacred Grove of Figs, Defender of the Little Church Down The Lane, son of Marquis Thomas Winfield III, Heir of King Charles the Thunderer, Lord Emperor of the Mizores, Regnant and Tribune to the Gracious Folk of the Plain, do hereby signify and confer with prince Elred of Gamberia. Greetings from the Keeper of the Holy Fan of Egypt….”

Pomposity-detection is as important a tool in link analysis as in political analysis. Truly powerful linking resources don’t have to look strong. They ARE strong. Their link profiles don’t have to extend to hundreds of thousands of links. They just need to include links from solid, reliable content across the Web.

If you want to know how strong someone else’s link profile is, you need to filter out all their duplicate links, their non-value passing links, their ghost links, and their filtered (Javascript or Nofollowed) links. What you’ll find after you clear away the dreck is pretty good at indicating what the real value of a site is.

Or you can just see how long it takes you to find some decent linking pages. If you have to dig deep through a site’s link profile for compelling links, there is something suspicious about the link profile. The way you can use Google’s link: operator is to gauge a site’s Google profile against other sites’ Google profiles.

That is, if Google’s random selection of links for your site provides fewer than 100 references and its random selection of links for your competitor’s site provides more than 1000 references, they may have a stronger link profile than yours. But you need to look at other factors (besides their Toolbar PR or their Yahoo! backlink report) to see what is really going on.

A lot of sites beef up their backlink profiles through internal links. While there is every reason to build as tight and powerful internal linkage as you possibly can, the acid test of external linkage is who links to whom. If two sites both have extensive link profiles, the site with the greater number of influencing links has the better profile.

An influencing link is embedded in relevant copy on a site that attracts a substantial amount of traffic from an audience interested in the topic. That is, a link from a Britney Spears site is better for your Britney Spears site if the other site happens to get a LOT of visitors. If your competitor has more Britney Spears links from high traffic Britney Spears sites than you do, your competitor has the better Britney Spears link profile.

Which brings us to my closing point: link profiles can be (and should be) managed from multiple perspectives. You need to look at link profiles by topic, by audience, by region (where applicable), by timeframe, and by incongruity. Incongruous links can be good as long as they make sense. If the incongruity doesn’t make sense, then it probably doesn’t help.

We’ll come back to the many faces of link profiles another day.