RocketCrawler logotype  RocketCrawler
 36
2025-02-05 | Time to read: 3 minutes
Facebook
Author: Chris Long

This analysis of 1M queries found that links still matter

Data showed links were most important for high volume, local and informational queries.

This was a solid data study from Patrick Stox from Ahrefs on the importance of links in 2025.

The study sought to ask the question "Do links still matter for SEO?".

They analyzed 1M queries across top 20 rankings for the correlations to different variables such as backlinks, followed links, referring domains, internal links and more.

Here's what he found:
1. In general, the study found that more links still correlated with better rankings. Data showed that both the number of referring domains (0.255) and the number of backlinks (0.248) had the strongest correlations to rankings of any other variable in the study.
2. When segmenting down the data by search volume, Patrick found that links matter even more for high search volume queries. While there was a 0.2+ correlation for queries with 5K volume, it spiked to 0.3+ for searches with 100K+ in volume.
3. Overall, links correlated with rankings slightly less than the last study. In their last study in 2019, links had a 0.27 correlation with rankings. In this study it was between 0.22–0.24.
4. Interestingly, link mattered for local queries a lot. Both referring domains and backlinks had relatively high correlations of 0.33 for local rankings.
5. The study also broke down the correlation by search intent. The data found that for Informational searches, referring domains (0.278) and backlinks (0.269) were most correlated.
6. Surprisingly, links with Commercial intent had the lowest correlation for both referring domains and backlinks with approximately 0.2 for each. This might be because commercial queries are generally lower in volume compared to informational.
7. I was also a bit surprised to see that keywords in the URL had such as negligible correlation (0.034) to rankings. While it's not a huge factor, I thought it would be a bit stronger.

I know a lot of people are going to mention that correlation isn't causation, but I appreciate any study that actually brings data to the table.

Author: «Chris Long»

Rate this article
5/5
1



<< Back