• Register
• Log in

SEO Theory - SEO Theory and Analysis Blog

Algorithm analysis, Web community relationship analysis, SEO practices and techniques, industry news, etc.

• Home
• About SEO Theory
• SEO Glossary

Use snapshot metrics to measure optimization

Posted by Michael Martinez on September 6, 2007 in Competitive Analysis, SEO Metrics

How well optimized is your Web site? There is really no accurate metric for gauging the quality of a Web site’s optimization. There is certainly no universally recognized metric for measuring optimization.

Search engine optimization is “the art of designing or modifying Web pages to rank well in search engines”. The technical definition eliminates nearly all link-based search placement strategies and some people would argue that is an insufficient definition.

If we accept for the sake of discussion that links are nearly as important as Web page content for search engine optimization, we can test some metrics for optimization that are relatively fair to both the link-based marketers and the content-based marketers. But it’s not possible to design a metric that shows whether links matter more than content (or vice versa).

One way to evaluate optimization success is to look at how many queries for which a site ranks. However, large content sites generate many more search referrals than small content sites so it’s hardly fair to say that a 1-million page site is better optimized than a 10-page site.

Another way to measure optimization success is to compare raw search referrals, but again large content sites will usually outperform small content sites (although conceivably a site with 400,000 backlinks may draw many thousands of search referrals on a monthly basis regardless of how small it is).

But it’s not possible to look at most other sites’ referral data so you can’t really build an unbiased metric based on search referrals.

Search visibility, however, is something that can be measured in a limited fashion. You can build a list of queries for a vertical and then compare rankings across multiple search engines. Meta search tools like IXQuick are good for such analysis.

IXQuick, especially, offers valuable insight into a Web site’s search visibility because it has a ranking algorithm that generates from 1 to 5 stars, indicating how many search engines list a given URL in the top ten results.

So while scanning IXQuick for a set of 30-40 queries may not be efficient, it can help you build a visibility graph on which you plot listings for a given site. I don’t know of any SEO tools that do this but it would be a very nice tool to have. If you could compare two or more Web sites’ visibility graphs, you’d get a pretty clear idea of which sites have the stronger visibility.

Search visibility is one indicator of optimization but it’s not a very strong indicator. Another way you can use IXQuick to evaluate a Web site’s optimization is to run a site: query. IXQuick will sort URLs by how high they appear across its search base. Relatively few Web sites earn more than 3 stars for even their root URL. So if a site has 2 or more 3-star or better listings, it has pretty search visibility.

A Web site with more than 25 top ten results may not have very good internal linkage. As a rule of thumb, if a site search in IXQuick returns fewer than 25 results you’re probably looking at a site with very strong internal linkage. If the site has 30 or more results its internal linkage is probably weak.

You could pretty easily define an IXQuick Strength check by assigning 1 point for every star a site is awarded in a query result. Internal link strength would have to be calculated differently, however, as the fewer pages a site has in the site query the better, so I would go with an average star rating for internal link strength.

SEO Theory would rate a 1.2 for internal strength. We have 25 listings and a total of 30 stars.

Searchengineland, by comparison, would rate a 1.5 for internal strength because it has only 20 listings but also earns 30 stars.

Searchengine watch rates a 1.1 with 29 listings and 32 stars.

CNN rates 1.34 with 29 listings and 39 stars.

Xenite.Org rates a 1.3 with 23 listings and 30 stars.

How well does your site measure up? Do you rate as poorly as Geocities, which has 40 1-star listings? That equates to a strength of 1 (the lowest possible rating).

Matt Cutts almost gets a 1.2 (his score is 1.1923…).

One parked domain earned a total of 1 listing with a single star. So he gets an internal link strength rating of 1, which shows that this metric is far from perfect.

You can also use Dogpile for site queries but Dogpile’s rather bland results don’t provide much opportunity for flexible metrics. On the other hand, they do tend to show more listings than IXQuick. Xenite has 53 listings on Dogpile. SEO-Theory has 54 listings. Searchengineland has 56 listings.

We could compare our IXQuick and Dogpile visibility with a ratio. For example, Xenite has 2.3 listings on Dogpile for every listing on IXQuick. Searchengineland has 2.8 listings on Dogpile for every listing on IXQuick. Matt Cutts has 2.23 listings on Dogpile for every listing on IXQuick.

These SEO-based samplings are too few to establish statistical validity but there does appear to be a correlation between the IXQuick internal link strength valuation and the IXQuick-to-Dogpile listing ratio. Assuming a large pool of similar comparisons showed (say from 200 to 1000) a similar correlation, we could conclude that these ratios indicate how well optimized certain sites are.

Danny Sullivan’s a pretty smart guy who has been reporting on the search engine optimization industry for years, and he has a team of crack SEOs working with him. I don’t think anyone should be surprised to learn that Searchengineland performs well in these metrics.

Nor would I hold it against Matt Cutts for earning deceptively low rankings. He is a non-SEO who has made a very big mark in an SEO industry. Matt has brand value that these metrics don’t even attempt to measure.

Search visibility snapshots, like all other SEO metrics, won’t reveal the whole picture to you. But you can form a pretty quick opinion of how much search coverage a site has by looking at these types of metrics.

You might be surprised to learn, for example, that Wikipedia’s Engliah-language section only rates an internal strength of 1 on IXQuick. But they also have 2.77 listings on Dogpile for every listing on IXQuick. This apparent disparity in ratings underscores just how important content is to long tail optimization. The more content you have, the more top ten listings you’re likely to earn.

But I think the IXQuick rating is accurate enough when compared to the other sample sites I show above. After all, Wikipedia doesn’t really optimize its content, despite the tight interlinkage and heavy repetition that some of its popular articles enjoy. Those articles are the tip of the iceberg, but the rest of the berg is starved for both content and strong internal links.

Another site that corroborates my Wikipedia test is About.com. It rates an IXQuick Internal Strength of 1.1428… but has only 1.8 Dogpile listings for every IXQuick listing. That is, About.com doesn’t have very strong internal linkage. Hence, more of its pages are likely to grab top ten listings.

Why should a large content site with weak internal linkage grab more top ten listings across multiple search engines than, say, a mid-sized content site with strong internal linkage?

The answer, I feel, is that strong internal linkage limits your search visibility. You’re focusing on too few pages to achieve maximum search visibility.

However, maximizing your search visibility may not be the best path to choose unless you have a lot of content. There has to be a statistical flip point where a small to mid-size site with limited search visibility but a few good strong rankings performs about as well as a large content site with few to no strong rankings.

We’ll call this the search visibility equilibrium point where the weight of content and the weight of links cancel each other out. On the small site side, linkage helps you more because you just don’t have enough content to attract large volumes of traffic. On the large site side, linkage matters less because you can spread your content far and wide.

Hence, a large content site that builds more top ten rankings than average should rock. And I believe that Searchengineland has done a pretty good job of building traffic and visibility over the past few months. These metrics are good quick evaluation tools because they don’t promise too much or deliver too little.

By themselves these metrics don’t mean much, but as you collect data for more sites you’ll see that any anomalies in the general patterns will usually flag significant optimization (or significant lack of optimization).

You should be able to map the strengths and visibility of 20-50 sites for any vertical and see who the strongest players are. If you’re intimately familiar with a vertical you may see a few awkward ratings and ratios but I think you’ll find yourself saying, “Yes, I knew that. I just couldn’t picture it before.”

These are not high definition metrics. You cannot zoom in on specific indicators. But with some work and innovation we may be able to refine these metrics and come up with some interesting SEO tools that tell us something more useful than what today’s current batch of SEO tools provide.