Does the size of the featured image or its “dimensions” affect Google’s organic search rankings? That’s something that I decided to put to the test. This is where I should tell you to read on to find out the shocking truth, except it’s not that shocking for reasons you’ll soon understand.
Disclaimer: The sample size used in this test is below statistical significance, and I would advise you to run similar test(s) on your own website before drawing firm conclusions. The purpose of this experiment is to provide a starting point for further analysis, and I would love to hear about your own findings.
Hypothesis
The old best practice for images was to set them to the max-width of the largest breakpoint for a web page’s content. In other words, a page with a 1000px wide layout would have no reasonable use for an image with a width of 1200px. This made a lot of sense and helped your pages load faster.
Things have changed over time because of responsive web design and the need to provide content on other platforms such as Facebook, Twitter, Pocket, Google Discover, and more.
The web evolved, so technology followed. Today, you can upload large images to a content management system (CMS) like WordPress, creating multiple images that can be served dynamically. We can use <img srcset> and <picture> to specify a group of images for responsive content. Improvements in image file formats, compression, and hardware enable us to provide more images than ever before and load them faster than ever before.
Google’s own documentation for Google Discover recommends using large images that should be at least 1200px wide. However, Google Search Central’s image best practices guide has no recommendations for dimensions.
We know that images are important, and it would appear that image dimensions are important, but without the recommendations around Google Search, it’s all hearsay. Common sense would suggest that using larger images, especially featured images that are used across many platforms, would be a consideration for gauging higher-quality content in a bid system.
So my hypothesis was to replace small featured images with large 1800x1800px images and then monitor the result to confirm whether this could be an image-based ranking factor.
Method
- Select 15 articles with stable rankings that had small(er) featured images for the test group.
- Select another 15 articles like the above and use them as the control group.
- The images would stay the same as before, use the same filenames with a single letter added, the same alt text, metadata, etc.
- The images were NOT unique (for those interested in that sort of thing)
- The featured images were NOT displayed on the front end to minimize page experience interference in the tests.
- Test the URLs individually and utilize a changelog in Google Sheets to monitor progress.
All that I’m looking for in this test is movement, whether positive or negative, to indicate that the featured image size could be a possible ranking factor.
An improvement, decline, or null result would be measured through a variance or a “deviation” in historical data. Not the most accurate in the world, but good enough for this test.
Findings
The findings were quite conclusive regarding how switching to a larger featured image precipitates ranking fluctuations.
After reprocessing the page, the average position for this article’s main keyword started to fluctuate more.
There were also pages where impressions began to climb quite rapidly, owing to some keyword positions climbing more than others, including minor improvements on the target keywords.
Test Group
| Improved | Declined | Null | Total movement % |
|---|---|---|---|
| 7 | 7 | 1 | 93.33% |
Interestingly, this group experienced mixed results from adding a larger featured image. However, some of these pages are still experiencing higher-than-usual volatility since the change. This could be seen as another sign that featured image dimensions can affect rankings, or it could be coincidental.
Control Group
| Improved | Declined | Null | Total movement % |
|---|---|---|---|
| 3 | 4 | 8 | 46.66% |
There was a big difference between the control and test groups. The test group experienced movement in 93% of updated pages versus the 46.6% in the test group that were left as is. The movement correlation is quite strong here, but it could be coincidental.
Conclusion
It appears that the size of featured images could be a possible small ranking factor. However, the data wasn’t conclusive enough to confirm this. This is possibly due to the small sample size or the inability to eliminate enough variables from the tests. There’s certainly some correlation there, and to me, that’s enough of a reason to continue using larger image sizes when we consider what we know about why that’s important in the first place.