SEO research & experiments

Experiments, studies and columns. Most of it lives on ovisy.hr, and this is the full list.

1. Do exact match domains still work? I spent $224 to find out Experiment

   ovisy.hr/blog/do-exact-match-domains-still-work · 2 Nov 2026

   7,374 US/UK keywords, 106,130 search results and ~60,000 LLM answers. An EMD won't rank you by itself: only 4.7% of new EMDs reached the top 10, vs 5.7% of other new domains. But LLMs picked the EMD 59% of the time where 20% was expected. (Yes, this site's domain is an EMD, I'm aware. )

   Microsoft Excel 97 - EMD_TOP10.XLS

   New domains that reached the top 10

   New EMDs4.7%

   Other new domains5.7%

   0%4%8%

   Source: EMD experiment, 7,374 keywords

   Microsoft Excel 97 - EMD_LLM.XLS

   How often LLMs picked the EMD

   Picked the EMD59%

   Expected by chance20%

   0%40%80%

   Source: ~60,000 LLM answers

2. The SEO checklist, graded by evidence Tool

   ovisy.hr/seo-checklist · 28 Sep 2026

   161 tactics graded from A to X, and only 13 get an A. The median controlled A/B test moves the metric +16%, while uncontrolled case studies claim +106%. Funny how that works.

3. Treba li nam još uvijek SEO? Proučio sam 9.249 dokumenata da provjerim Column

   www.netokracija.com/seo-kresimir-corluka-kolumna-istrazivanje-247020 · 22 Sep 2026 · Netokracija, in Croatian

   Do we still need SEO? I read 9,249 documents to check. My column for Netokracija.

4. The SEO Meta Study Study

   ovisy.hr/blog/seo-meta-study · 10 Sep 2026

   30,000+ documents screened down to 9,249 sources and 161 tactics, each graded by how good the evidence behind it is. Formal record on Zenodo (DOI 10.5281/zenodo.22768160). P.S. I'm not really a scientist lmao.

5. Does llms.txt do anything? Eeeehh.... Finding

   kresimircorluka.com/research/ · Oct 2026

   Two entries in the entire literature. One pairs llms.txt with schema markup, the other reports +25% organic sessions, and neither tests the file on its own. Cheap to add, don't expect miracles. (This site has one, because it was cheap.)

6. Title tags still work (and can still hurt) Finding

   kresimircorluka.com/research/ · Sep 2026

   11 controlled title-tag tests: 7 wins, 4 losses. I'd filed title tags under boring basics for years, and then the data disagreed.

7. Only 11 of 572 tactic entries with a measured effect went negative Finding

   kresimircorluka.com/research/ · Sep 2026

   9 from SearchPilot, 1 from Moz, 1 from Ahrefs. Great job, SearchPilot. We'd all learn more if more people published their flops.

8. Rebrands and migrations: 523 days to recover Workshop

   kresimircorluka.com/research/ · Jun 2026

   Average recovery time after moving to a new domain in the data I reviewed, and 17% still hadn't recovered when the data ended. Plan the migration before the logo.

9. Why do Croatian news portals ignore SEO? Analysis

   kresimircorluka.com/research/ · Feb 2025

   I expected big evergreen libraries pulling massive traffic, but mostly found accidental rankings. There's a lot of traffic lying around for whoever plans for it.

10. Lessons learned in 200+ SEO audits Workshop

    kresimircorluka.com/research/ · Jun 2024

    The problems I keep finding on real sites: categorisation, JavaScript, internal links, and the classic robots.txt pushed from dev to production.

11. 6 common SEO problems in Croatian webshops Analysis

    kresimircorluka.com/research/ · Feb 2024

    The ones I actually see in Croatia, starting with that robots.txt again.