He Dug Through 9,249 SEO Case Studies. Here's What Actually Works

E1207: An SEO spent his summer collecting more than 30,000 documents to find which SEO tactics work. That SEO, Krešimir Ćorluka, narrowed them to 9,249 studies, tests, and papers, then ranked the strongest evidence.

I covered some of his most surprising findings in an earlier episode. This time, Krešimir joins me to explain how he built the research, which results he trusts, which ones he questions, and which tactics he prioritizes in his own SEO work.

We dig into:

- How he used Ahrefs, Screaming Frog, two AI models, and manual checks to sort thousands of SEO studies.

- Why so many SEO case studies fail to prove causation: no control group, multiple changes at once, seasonality, and agencies rarely publishing failed experiments.

- Why adding "Just Updated" to a title tag beats putting the current year in it.

- What the tests found about exact-match internal link anchor text versus using lots of variations.

- Why dynamic and static prices in title tags produced different traffic outcomes.

- The Google Lens test that reported a 90% increase after adding EXIF location data and ImageObject schema, and how other EXIF tests found no benefit.

- How removing review-related schema coincided with rankings recovering for two of Krešimir's clients.

- Why he questions many AI search optimization studies, especially those measuring ChatGPT mentions from a handful of prompts.

- What the evidence says about LLMs.txt, expert quotes, statistics, and tables for AI citations.