New research data shows the inbound link patterns of sites that rank. The results of the study has mixed results. Looked at one way, the research indicates that links tend to play less of a role in ranking. Interestingly, there were startling differences when commercial queries were compared with informational queries.
In the following graphs, the higher the number score, the higher the correlation of backlinks are to ranking. Higher is more correlation. Lower is less correlation.
No Change in Link Influence
The research compared the same 6,000 queries over time. From 2016 to 2018, the researchers sampled the data and noted how many links were contained by the sites ranked at the top of Google.
The idea was that if links played a major role or less role in the search results, the backlink information would reflect the changes.
A comparison of the 6,000 queries showed there was no statistical difference in link influence from 2016 to 2018. The evidence clearly shows that there was no change in how Google treats links.
In a comparison of 6,000 search queries, researchers discovered there was no change in the number of links contained by top ranked sites.
The researchers then increased the data set to 16,000 search queries
The backlink amounts for the top ranked sites across 16,000 search queries showed that, statistically, there was no difference
Perhaps this is a reflection that an algorithm doesn’t change much in two years? Then a surprising thing happened.
The researchers increased the size of data sample from 16,000 to 27,000 search queries. What they discovered was like night and day between the two data sets.
Commercial Versus Informational Queries
The research expanded to 27,000 search queries and that’s when differences began to be noticed.
The researchers split the queries between informational and commercial queries.
The sites ranking for informational queries contained more backlinks than the sites ranking for commercial search queries.
Why Do Informational Sites Contain More Backlinks?
It’s counterintuitive that less links were powering commercial search queries. Internet Marketing implies that something is being sold. So why do the sites that rank for commercial queries containing less links?
The researchers then split the commercial search queries by niche sectors and noted the differences.
All others .232
All Sectors Together .238
As you can see in the list above, the top ranked financial search queries contained more backlinks than all the other sectors. Commercial niches outside of the three named sectors contained the least amount of backlinks.
All others .320
All Sectors Together .314
Interestingly, the backlinks of top ranked informational sites were ranked in different proportions than the commercial sites. Sites ranked for Medical queries contained more backlinks than sites in the financial sector.
Links Matter Less? Depends on the Data
Here is where the data shows a different perspective on whether or not links matter in the search results. Starting with the first study, they normalized all the backlink information by ranking position. Here is how Eric Enge explained it:
“We normalized the total links for each individual query for all the results from 1 to 100. Then we summed all the normalized link totals for the number one results, separately summed all the normalized link totals for the number two results, and so forth. Then we ran the Spearman Correlation calculation on that.
So they took position 1 of all 6,000 search queries and normalized the backlink data. Then they did it to positions 2 through 100. Then they ran the calculations. They repeated this process on the 16,000 query and 27,000 query data set. The results were extraordinary.
Analysis by Ranking Position of 6,000 Queries
The data shows that links mattered more in 2016, then mattered less in 2017, then mattered more in 2018, but less than in 2016.
The backlink correlation of sites ranked in 6,000 queries by ranking position.
Analysis by Ranking Position of 16,000 Queries
The results for 16,000 queries are different than the results for 6,000 search queries.
The data shows that links mattered a lot in all three years (2016, 2017, and 2018). But this time it shows that links mattered more in 2017 than it did in 2018.
The backlink correlation of sites ranked in 16,000 queries by ranking position.
Analysis by Ranking Position of 27,000 Queries
The analysis of 27,000 queries shows that in 2018, links matter more than in the other data sets of 6,000 and 16,000 search queries.
This analysis shows that links matter more than ever for ranking:
The backlink correlation of sites ranked in 27,000 queries, normalized by ranking position.
The 16,000 query data set shows that in 2018 links mattered less than they did in 2016 and 2017.
August 2016 .815
May 2017 .864
August 2018 .770 ← Lower correlation
That’s a downward trend.
The 6,000 query data set showed an up and down pattern, with May 2016 showing that links mattered more than in 2018
May 2016 .951
August 2016 .866
May 2017 .802
2018 .911 ← 2018 scores less than May 2016
The researchers did not have the 27,000 query data for 2017 and 2016. So we can’t really know if link influence score on the 27k data set is part of a downward trend.
It seems that this method shows an extraordinary correlation between links and ranking. The difference between this method and the other methods is profound.
The question is, which method is the most accurate?
Is the importance of links going up and down depending on the month or year?
Or is the importance of links going steadily down?
To me it seems there is no clear answer.
The authors of this research smartly observe the difficulty of using statistics to identify clear correlations:
“…relevancy and quality are very large ranking factors, as they should be. In addition, there are many other factors such as Google’s need to show diversity in the SERPs…”
Put another way, there are other factors that step in after the links and pages are ranked. Factors such as geography, whether a search is local, informational or commercial will override the winner in a link ranking contest.
The fact that Eric Enge is associated with this research makes it important because Eric and the Stone Temple agency are aware of the limits of data mining and that knowledge is reflected in their report.
The value of this research, in my opinion, is to highlight different patterns of linking between successful commercial and informational sites. But drawing conclusions from those patterns can be risky.
For example, the top ranked informational medical queries contained more backlinks than the commercial medical queries.
Informational medical queries .337
Commercial medical queries .292
The data does not mean that medical informational sites need more links than commercial medical topics to compete. It may simply reflect an overall smaller link community in competition for commercial medical terms.
For example, the competition for queries related to local hospital procedures may be lower. The above statistics may also reflect that informational medical queries are more popularly cited by news, blogs and other web publications.
So one possible way to look at the data is that the link community associated with informational medical communities is larger than the link community associated with the commercial search queries. But we can’t say for certain unless we know the relative sizes of both communities.
Any observations based on the limited data are inconclusive.
A recent article in Wired titled The Exaggerated Promise of Data Mining had this to say about mining statistics for meaning:
“Finding something unusual or surprising after it has already occurred is neither unusual nor surprising. Patterns are sure to be found, and are likely to be misleading, absurd, or worse.
In 2011, Google created an artificial intelligence program called Google Flu that used search queries to predict flu outbreaks. Google’s data-mining program looked at 50 million search queries and identified the 45 that were the most closely correlated with the incidence of flu. …After issuing its report, Google Flu overestimated the number of flu cases for 100 of the next 108 weeks, by an average of nearly 100 percent. “
What is the Status of Links and Ranking?
There are many clues but the nature of data mining precludes from reaching definitive answers. One thing we can be certain of however is that links are still a part of the ranking algorithm.
It has been many years though since links have been the deciding factor for ranking. The value of the research is that it gives a high level view of the state of links. And that’s not a view we are often privileged to see.
Read the entire report at Stone Temple.
Images by Shutterstock, Modified by AuthorScreenshots by Author, Modified by Author
Subscribe to SEJ
Get our daily newsletter from SEJ’s Founder Loren Baker about the latest news in the industry!