SearchMetrics just released an infographic on search ranking factors, and it’s a bunch of bunk. I won’t dignify them by linking to their site, but here’s an Imgur link if you want to lose some brain cells.
Where to begin with this mess?
First of all, the title of the infographic is misleading. “Ranking Factors” suggests that the data in the infographic actually, you know, has an impact on rankings. But that’s not the case. SearchMetrics just did a correlation analysis of some subset of websites, and put those correlations into a pretty picture. That’s like analyzing a bunch of websites and finding that the highest ranking sites mostly had black text, and changing your text color based on that “ranking factor”. It doesn’t make sense.
Take for example social signals. One might conclude by looking at all the pretty bars that social signals have a high impact on rankings. However, that’s not the case. Google has said repeatedly that they do not take social signals into account for rankings (except Google+, which impacts logged in users).
Or take a look at their claim that having keywords in the URL does not correlate well with rankings, which clearly defies my experience as a professional SEO. Even if the keywords did not have a direct impact on rankings (which they do, even post-EMD update) having the keywords in the URL still helps rankings from a backlinking perspective.
Perhaps most damning of all, almost none of the correlations in the survey have any statistical significance. Here is a guideline for interpreting Spearman’s correlation coefficients:
- 0.3 – 0.4.9: Correlation is moderate to low
- 0.16 – 0.29: Correlation is weak to low
- Below 0.16: Correlation is too low to be meaningful
Going by this rubric, only 6 of their correlations had a moderate to low correlation, and the majority of the data points that they list have a correlation which is “too low to be meaningful”. That’s the entire Content and Technical section, which make up half the infographic!
Anyway, I won’t waste any more words analyzing this infographic, but it’s actually being published on a few internet marketing blogs, so I wanted to clearly disprove these so-called “ranking factors”.
If you actually want actionable ranking factors, Moz puts out a great list every year based on surveys of the top SEO professionals in the field. I also recently came across this set of ranking factors from Netmark.com, which line up closely with my own personal experience doing SEO.
3 thoughts on “SearchMetrics’ 2013 SEO Ranking Factors Are Useless”
Takeshi, Interesting article. I saw the correlation study release today, but didn’t get a good look at it until just barely. Your insights align with mine. I don’t think what Searchmetrics reported was bad, but I don’t think it was super great. I appreciate the kind words about Netmark. We worked EXTRA hard for quite some time to come up with the data that we did. Keep on rock’n 🙂
Takeshi, the info graphic is based on a 70 pages whitepaper. You only mentioned the info graphic which makes absolutely no sense. You didn’t mentioned how the data was aggregated etc. You put much time in writing this article without reading the whitepaper. With 6+ years SEO experience… I can recommend again reading the white paper, because it is explained as well why a high correlation maybe is not a high ranking factor and why a neutral correlation could be a high one. And in the end data does not lie. In my mind much better than a survey of a few SEOs.
Marcus, I think it’s fair to criticize the infographic since that’s what’s being passed around the web. I doubt very many people who saw the infographic took the time to read the full 70-page whitepaper. And that’s the point, that the infographic is misleading. It calls itself a “ranking factors” study, and yet the information presented is just a weak set of correlations. I would much rather trust the opinions of the world’s top SEOs about what’s working for them, than this infographic. Data may not lie, but if you interpret it incorrectly, it may as well be a lie.
SEO Theory just posted a good article on this subject as well.