A few months ago we undertook a small study to investigate if the anchor text of incoming links still had an observable influence on the ability of a page to rank higher in Google.
That study was very well received in the SEO community, but the the sample size was very-very small – only 50 keywords.
So we felt obliged to reproduce this study on a (much) larger scale to see if the takeaways would remain the same.
Sample Size
This time we took a sample size of 16,000 keywords and for each of these keywords we analyzed Google’s top 20 search results.
That’s 320,000 pages in total.
For each page (URL + Keyword) we collected the following metrics:
- Ahrefs rank of the target URL
- Total # of referring domains of the target URL
- # of partial keyword match referring domains
- # of exact keyword match referring domains
- # of DoFollow partial keyword match referring domains
- # of NoFollow partial keyword match referring domains
- # of DoFollow exact keyword match referring domains
- # of NoFollow exact keyword match referring domains
- Avg. DR of the partial keyword match referring domains
- Avg. DR of the exact keyword match referring domains
Experiment #1
First of all, we wanted to find out if there was any correlation between the percentage of exact-match anchor text links and partial-match anchor-text links in a page’s backlink profile and it’s ability to rank higher.
Here’s what we got:

From this graph it’s clear that pages that rank higher tend to have a bigger percentage of exact-match anchor text links in their backlink profile.
Surprisingly, the correlation of partial-match anchor text links wasn’t too prominent.
But that was a percentage of links from the total backlink profile of a page.
What if we look at a raw number of referring domains with both exact and partial match anchor text links?

The takeaway is clear – anchor text links from a variety of different domains correlate with higher rankings in Google.
Experiment #2
One thing that Experiment #1 revealed is that pages with more backlinks from different domains tend to easily outperform pages with less backlinks. This is what we would expect to see.
So in Experiment #2 we decided to level the effect of a strong backlink profile and only compare the data on pages with similar numbers and quality of incoming links.
Thus we have chosen the keywords where the Top 5 Google results’ URL Rank had standard deviation of less than 30% of their average value. This resulted in about 2,000 keywords (out of 16,000).
Here’s how this changed the numbers:

As you can clearly see, in the case where pages had equally strong backlink profiles it was the % of partial match anchor text links that mattered the most, while the % of exact match links stayed more or less consistent across the Top 5 results.
Takeaways
Similar to our previous study the main takeaway is this:
But even though we saw a very strong correlation, we should not forget that this is just one of the many factors that Google takes into account when ranking pages.
By leveling the effect of a strong backlink profile we got totally different correlation numbers.
Which clearly illustrates that focusing on a single ranking factor will get you nowhere.
As always, any questions or comments please leave them below 