Correlation vs. Causation in SEO
Correlation vs. Causation
Correlation describes the relationship between two or more variables that have a measurable effect on each other. In SEO, we look for variables that have an impact on search engine rankings. By studying the features of top ranking results versus those of lower ranking results, we can determine if there is a correlation in Google’s ranking algorithm. For example, we may see that pages with higher search engine rankings have a more diverse collection of inbound links.
However, correlation does not imply causation. Just because higher ranking pages have a given feature that is lacking in lower ranking pages does not mean the feature was the cause. In SEO, the relationship varies with each webpage and there are other variables that may contribute to a correlation. For example, we may see a correlation between high ranking webpages that have compelling content, optimized title tags, and a diverse collection of backlinks but since we do not know the exact value Google places on these factors, we cannot attribute them to causing the page to rank higher. For these reasons we say that most SEO is correlative instead of causative.
Below is nice infographic created by Moz, portraying some basic examples of correlation and how they differ from causation.
Correlation Data for SEO and Social Media Analysis – Part 1 – Whiteboard Friday
In part one of this two-part video series Rand Fishkin, founder of SEOmoz, talks about the general importance of correlation data and how we can use it for research in our SEO and Social Media campaigns. For example, by looking at correlation data, you may notice that “keywords in the alt attribute in an image, predict higher average correlation than keywords in H1.”
Correlation Data for SEO and Social Media Analysis – Part 2 – Whiteboard Friday
In part two, Rand talks about specific correlations between SEO tactics and their effects on rankings. In a large analysis performed in June of 2010, SEOMoz looked at different ranking factors and correlations across eleven thousand search results in both Google and Bing. Some of their findings included:
- There was not much difference in the correlations of search result rankings between Google and Bing. So the results with higher rankings in Google had the same or similar features to the results with higher rankings in Bing.
- Facebook “shares” are positively correlated with Google rankings.
- “Nofollow” links have a positive correlation with rankings.
- The percentage of “follow” links has a negative correlation with search engine rankings. Suggests that webpages or websites that have more nofollow links perform better than those that have less nofollow links. The reason here could be that it is more “natural” to have nofollow links pointing to a site – suggesting that the site is more interactive within it’s “ecosphere” (blogs, Facebook, Wikipedia, etc).