New "Perfect Page" Report

What are the elements that make up a "perfect" page to rank well with a crawler-based search engine? This is often the holy grail many site owners seek. I've written before about resources and programs that attempt to help you find the perfect page. Now there's a new one, the "Search Engine Ranking" study from Axandra.

The company analyzed over 100,000 pages that ranked in the top ten at Google, AltaVista, Inktomi (using iWon results), AllTheWeb, Wisenut and Teoma. It then further broke them down by page elements and other factors, with the goal designed to help you understand which criteria seem most important for ranking success.

I skimmed through one of the reports, for Google. It was based on searching for nearly 2,000 queries on Google in April 2002 and analyzing the pages listed in the first page of results for those queries. The queries are all shown at the end of the report, ranging from "abstract art" to "college basketball" to "vegetarian diet."

There's no lack of data. The report reveals such things as 77 percent of the top ranked pages in Google are the root page of a web site, while only one .nsf page made it into the top ten. Of course, there are a lot more root pages on the web, so its far more likely that the odds are there will be more root pages ranking well, rather than there being ranking discrimination against the rather uncommon .nsf extension.

The report goes on to show other factors such as the overall size of documents most commonly ranked well in Google, amount of visible text, number of search terms found in high ranking pages that appear in bold text (not a lot), position of search terms on high ranking pages, HTTP response codes of pages, age of domain registration (top position does tend to go to older domains, but not always), presence of Flash, Java, number of meta tags, use of the meta refresh tag (over 90 percent of top ranked pages don't have it).

Some stats are absolutely fascinating. For example, the biggest chunk of high ranking pages at Google for a particular query -- about 35 to 40 percent -- did not contain the query terms at all.

There's link analysis, hard at work -- and a real blow to all those who spent tons of time trying to calculate the best "keyword density" for their pages. Then again, the report also found that by far the biggest chunk of top ranked pages at Google -- about 30 to 40 percent -- had no external links pointing to them at all.

There's a great little chart to measure the importance of directory links, with the DLY acronym used to indicate a link from The Open Directory (D for Dmoz), LookSmart and Yahoo. The chart seems to support that the popular wisdom that getting links from all three directories is crucial, given that the vast majority of top ranked pages are DLYs, as opposed to just DLs or DYs.

Overall, I see this report more as a great myth buster than a recipe to success. For instance, the vast majority of top ranking pages are not using hyphenated URLs, something often done in the belief that by embedding keywords into a domain name or URL string, you'll rank better for those terms.

There might be a few secret weapon elements the report has found, but you'll probably be lost in the over 300 pages of data trying to discover this. Indeed, the report's biggest weakness is that it lacks a good executive summary. You want a section where someone has read through all the data, then given you some guidance about what they think are the most important elements and why.

The other big problem is that crawlers change. What's listed today in an analysis could be completely different tomorrow. The authors acknowledge this issue in that they intend to do quarterly surveys, and first time purchasers of the existing reports are supposed to be given a significant discount on future releases.

Should you get it? If you are on a intense quest for the perfect page, the $50 price tag per engine will be cheap compared to the time to do the study yourself. You can also get a discount if you purchase a report for several engines bundled together. Even if you aren't after the perfect page, you might find the money well spent to get some answers about particular page aspects you wonder about.


Page Analyzers / "Perfect Page" Tools

A list of other tools and resources for those after the perfect page.

In Pursuit Of The Perfect Page
The Search Engine Update, Feb. 22, 2000

The points I make in this older article remain valid. Top ranking pages are not perfect, and indeed, attempts to get a perfect page may be the exact opposite of what you want to do in order to rank well.