SEOMOZ2013 Search engine Ranking algorithm analysis

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

Every two years Moz will do a ranking factor survey to determine which factors are most relevant to Google's rankings. The survey consists of two parts, one for SEO practitioners and one based on a large number of data dependencies.

Before delving into specific data, share 5 key findings of the year:

1, the relevance of the page weight in all the factors ranked highest;

2, social performance, especially Google + and Facebook share a very high correlation;

3, despite the implementation of Google Penguin algorithm, anchor text is as important as the relevance of the previous;

4, the first study of schema.org and structural data used in the ranking of relevance; (Schema.org is a series of words, through the site owner's tag, make the search engine easier to understand the content of the site. )

5, on the chain, keywords and accurate matching the domain name of the data collection more.

Questionnaires

A total of 120 people participated in the questionnaire. The questionnaire asked participants to rate the importance of the rankings from 1 to 10 for different factors, and then to calculate the average score of all questionnaires. The highest scoring factors averaged 7-8 points, while the lowest scoring factor averaged 4-6 points.

Correlation experiment

The method of calculating correlation is the same as that of 2011. First through Google AdWords to find a number of keywords (more than 14,000), this batch of keywords can cover different industries, different search volume. And then grabbed the top 50 natural rankings of the search engine from Google, all of which were crawled before Penguin 2.0 (June ago).

For each search result, all the factors that want to analyze are extracted, then the Spearman average correlation coefficient is computed by the data table.

When describing the results of a correlation, remember that correlation does not prove causal. (Correlation does not explain why this page is ranked first. The other one is in the second place; correlation tells us which factors are ranked in the previous page, and that the ranking in the back of the page does not; just because the top of the page has a ranking in the following page does not mean that this factor must be applied to the ranking algorithm. )

Link

Page-Level relevance: page weight, c-blocks, IPs, root domain, outside chain, to the page link site level of relevance: to level two domain name root domain, domain name weight, to domain name root domain, domain name mozrank, domain name Moztrust.

Page weight is a mozscape index which uses machine learning model to predict the ranking ability of links, and is the most relevant factor in experiment. As with the results of the 2011, there is a high correlation between the factors that derive diversity from the source of the link. In domain name and level two domain name, the correlation of level two domain name is bigger than the correlation of domain name.

In the questionnaire, Seoer also believes that links are very important.

The importance of the link in order: Link to the page links, links to the site link, link page relevance, Domain name Trust, page Trust, PR, link diversity, link feelings.

Anchor text

Over the past two years, Google has been hard hit by the overly optimized anchor text, but it is found from the data that the relevance of the anchor text is still very high, whether it is a partial match or a precise match.

Interestingly, the survey showed that the SEO practitioners believed that the natural anchor text distribution (reasonable brand words and the proportion of non-brand words) is more important than the number of links.

Includes: Natural anchor text distribution, partial matching links, accurate matching links.

The relevance of anchor text is the biggest difference between the whole experiment and searchmetrics. Speculation is that SEOmoz's experiments included keywords for navigation, and searchmetrics removed such words. Many navigation words are brand words, naturally there will be a lot of anchor text are brand words.

Page optimization

The relevance of documents and keywords was measured in two aspects of Word frequency score and language pattern score, and it was found that title tags, html content, meta tags, h1 tags were highly correlated.

SEO practitioners also agree that it is important to include keywords in headings and pages.

The experiment also tested the relevance of some additional page-optimized data, such as structural tagging (schema.org), and found that the correlation tended to be 0, so that these factors were not considered as ranking factors for the time being. Includes HTML characters, pictures, schema annotations, publisher callout for Google +, open chart callout, and author callout for Google +.

Precise/partial matching domain name

The discussion about EMD and PMD has been intense, and it seems that Google has been tweaking their algorithms. The experimental data show that EMD is 0.17, and. com's EMD is 0.2, similar to the 2011 results.

Surprisingly, the percentage of EMD domain names on the home page has declined year by year. In terms of relevance, EMD is a seesaw shape that rose again in the March of 13. The previous drop should have been Google's adjustment for EMD, and the rise in relevance suggests that EMD still exists in high ranking pages, but not so popular.

Social media

As in 2011, the data show that social media is very relevant, and then the questionnaire shows that SEO practitioners are not very optimistic about social media.

Social media can be transformed into other data to show dependencies, such as links, without becoming a direct ranking factor.

Summary

The weights of the factors that affect the rankings are as follows:

Link Weights for Web sites (20.56)

Link weights for Web pages (19.31)

Keywords and content of Web pages (14.87)

Other aspects of the Web page that have nothing to do with keywords (9.07)

Brand factors of the website (8.83)

User experience and traffic, search request data (8.28)

Social media factors for Web pages (7.28)

Keywords for website (6.74)

Other factors unrelated to the keywords on the website (5.26)

This concludes with three points:

1, link is still the most important algorithm. (Nearly 40%)

2, the key words on the page are the basis, many times more important than the link.

3, although the data show the relevance of socialization is very high, but practitioners are not very recognized.

SEO practitioners believe that the future of SEO will gradually discard the traditional ranking factors (such as anchor text, exactly match the domain name, etc.), but to in-depth study of the Web page for users, authors, structural data and social indicators of value.

This article by Shanghai SEO Company June Information collation provided, reproduced please specify the source: http://www.wenjuntech.com/sem/blog-1691.html