There’s been a lot of debate over the years about how to structure URLs to obtain maximum SEO benefit, highlighting the following considerations:
- Should the target keyword be included in the URL?
- How many subfolders should be used?
- How should the keywords be placed?
It’s a topic that our clients frequently ask about, so we decided to do some tests and answer the question: “Can we get any benefit from structuring a URL in a certain way?”.
The idea
Our original idea stemmed from a debate about the recommended URL structure for an ecommerce site. For example, if the target phrase was ‘maxi dresses’, would there be any danger of a URL being over-optimized by using a certain structure?
To find out the answer, we decided to test whether Google would favor subdirectories (keyword1/keyword2/) over a hyphen-connected keyword (/keyword1-keyword2/), effectively a keyword phrase.
The test
Here at Search Laboratory, we have our own portfolio of testing domains, operating in highly-controlled scenarios. These domains mean we can run tests on different aspects of SEO theory in order to submit only the best, proven recommendations to our clients.
Test process
- Select three different two-word keyword phrases (“keyword1 keyword2”) with zero results in Google
- Set up four test sites per keyword phrase, each with a page of content filled with ‘lorem ipsum’
- Structure each site with a different URL structure:
.com/keyword1/keyword2/
.com/keyword1-keyword2/
.com/keyword2/keyword1/
.com/keyword2-keyword1/
- Fetch as Google to get the pages indexed
- Allow enough time for Google to understand the sites
- Track each website’s KW ranking position in AWR.
Initial results
First set of websites
Structure ranking #1: .com/keyword1-keyword2/
Second set of websites
Structure ranking #1: .com/keyword1-keyword2/
Third set of websites
Structure ranking #1: .com/keyword1-keyword2/
Across all three keywords the #1 ranking page follows the structure:
.com/keyword1-keyword2/
The #2 ranking page follows:
.com/keyword1/keyword2/
Rankings for #3 and #4 are not consistent
We anticipated the two structures would carry the same weight from a ranking point of view, but they did not follow that pattern.
Takeaway
- Google favors the keyword phrase connected with a hyphen rather than split in to subfolders.
- Keyword order plays a role, Google favors the first keyword of the keyword phrase when placed early in the URL
We soon realized the way we structured the test meant that the middle sub-directory was not actually populated with content and was only there to build out the desired URL structure. In addition, the keyword phrase followed right after the root which might have affected our test.
To bring our test closer to a real-life scenario, and ensure the number of subfolders before the target keyword were the same for both structures, we conducted another test.
The second test
We started completely afresh with this test, but this time we only used two keyword sets instead of three, using the following URL structures:
.com/random/keyword1-keyword2/
.com/keyword1/keyword2/
.com/random/keyword2-keyword1/
.com/keyword2/keyword1/
The /random/ subfolder is named after a random keyword with zero results in Google and is totally unrelated to the keyword phrase.
Results
We had some issues indexing both sets of websites, even though we fetched as Google on a daily basis, so we let this test run for a few months. Google has two different versions of indexing, fast-track and normal indexing, which affects how fast a page is indexed and for how long it will stay in the index.
Not surprisingly, there was quite bit more movement this time around, with different structures competing for the first spot. After we saw consistent results, we were ready to compare them with the first test.
First set of websites
Structure ranking #1: .com/random/keyword1-keyword2/
Second set of websites
Structure ranking #1:.com/random/keyword1-keyword2/
The final classification of our test pages is as follows:
Conclusion
Comparing five keyword phrases with different URL structures – even after the inclusion of a subfolder to remove the root association – we can argue that the /keyword1-keyword2/ structure is the most likely to rank higher.
Next steps
We’re now looking to take our tests one step further and essentially create a real-life scenario site with content populated on every level. That way Google will have topical/entity association for the keyword phrases, which will help us test if a page with content ranks differently than one with keyword-in-URL alone. We will test for URL over-optimization and track both generic and longtail keywords.
If you have any questions about your URL structure, or any other technical SEO queries, please get in touch to speak to one of our experts.
Insights
