Study: Page Authority Hardly Matters for Comparisons

under SEO.

  • 4

Most SEOs spend a lot of time obsessing over a metric from Moz called Page Authority, and it's easy to see why. Nothing correlates better with rankings then Page Authority, which is built around a machine learning algorithm that incorporates the number of links pointing toward your site, trust in your site and the links to your site, and dozens of other factors.

According to Moz, Page Authority has a correlation with rankings of 28 percent. This is the most powerful metric we have. I'm a statistics major. A correlation of 28 percent is peanuts.

I decided I'd put together a quick study to emphasize just how small this effect really is on rankings.

How I did it

Using a random noun generator I chose 10 random keywords. I know, 10 isn't a big number. The point is, if I can't get anything useful out of 10 keywords, does it make sense for you to obsess over how high your Page Authority is for just one keyword?

Next, I chose two sets of 10 random numbers between 1 and 100 using this random integer generator.

I opened up an Incognito window in Chrome, went to Google, changed the settings to display 100 results per page, and searched for each keyword. Then I went through and checked the result corresponding to the random numbers chosen earlier. I looked up the Page Authority in Open Site Explorer.

Then I calculated the difference in rankings between the two results, and the difference in Page Authority.

Hopefully this image of my spreadsheet will help you understand what I did here:

page authority

Now, before I move on, take a look at the keyword "Wood." The page that was ranking 33 for that had a Page Authority of 35, and the page ranking in the 51st spot had a page authority of 71. That's a Page Authority difference of 36, and the one with the worse score is in the lead by 18 spots.

In fact, the difference in Page Authority had pretty much nothing to do with the difference in rankings. Take a look:

page authority chart

These are all over the board. When you find yourself thinking "I only need a Page Authority of X to beat this competitor," I want you to think of this graph.

Now, when I went ahead and ran a correlation test in Minitab, this is what I got:

page authority correlation

So, I got a correlation of 25 percent, and I'll take Moz's word for it as far as the actual figure being 28 percent. But what I really want you to realize here is that chart up there represents a correlation of 25 percent. Correlations this low tell you close to nothing.

That P-Value, what does it mean? It means that there is a 48% chance we would have gotten results at least this extreme if Page Authority had nothing to do with rankings.

That's right. After looking at 10 keywords and comparing 20 results, it's pretty much 50/50 whether or not anything's going on at all.

Again, I'll take Moz's word for it and assume that with their much larger datasets they've reached statistical significance.

What I want you to take away from this is that you can't look at the Page Authority of your competitors and say anything meaningful about how much Page Authority you need in order to beat them. There's much more to the algorithm.