Krešimir Ćorluka

Krešimir ĆorlukaSep 15, 2026

Napravio sam najveću SEO "meta studiju" ikad (valjda). 🔬 Krenuo sam s više od 30.000 dokumenata s weba. Tražio sam case studyje, split testove i akademske radove koji bi mi mogli reći što STVARNO radi, a što ne. I onda sam od toga napravio knjigu koja je besplatna (u komentaru je). Zašto je džabe objasnim na kraju.👇 Nakon filtriranja ostalo je nešto više od 9.000 dokumenata. Samo 1.391 je imao rezultat koji je netko stvarno izmjerio. Iz njih sam izvukao 7.329 nalaza, grupirao ih u 1.105 evidence clustera, a na kraju su 182 taktike zaslužile mjesto u ebooku. Od 769 taktika koje sam mogao ocijeniti, 78% nema niti jedan kontrolirani test iza sebe. Zato je u najvišu kategoriju ("REALLY LEGIT"), za koju trebaju barem dva neovisna kontrolirana testa, upalo samo 8 taktika. Ali, kužim zašto malo agencija radi kontrolirane testove. Kontrolirane testove je često teško raditi s klijentima, mogu biti skupi, a i vjerojatno agencije ne žele objaviti točne metodologije koje koriste. Ako objaviš da si klijentu povećao revenue 300% optimizacijom anchor texta, to je sasvim dovoljno dobar case study. Ne moraš konkurenciji objašnjavati točno kako si to napravio i što si testirao. Tako da, nažalost, ipak ne mogu reći da sam "RIJEŠIO SEO", jer jednostavno nema dovoljno kontroliranih testova da možemo sa sigurnošću reći što zaista radi. Ali isplati se baciti oko na rezultate studije. Samo vas molim da ne uzmete knjigu, ubacite je u Claude i kažete "de primijeni sve ovo na moj web kthxbye" jer stvari ne funkcioniraju tako. Rezultati dolaze s različitih webova, tržišta, industrija i tipova stranica i za dosta toga treba razumjeti kontekst prije nego što išta dirate. Ako želite ebook, samo napišite "ENGAGEMENT BAIT" u komentar i poslat ću vam link. jk jk Link je u prvom komentaru. Moram biti malo cheeky. 😶 P.S. Neki su me pokušali uvjeriti da ovo ne dam besplatno jer je metodologija moja, a u cijelu stvar sam utresao brdo novca i više od dva mjeseca rada. Realno, mogao sam naplatiti. Ali nije mi baš fer prema svim agencijama, firmama i ljudima koji su sve ove podatke i case studyje objavili potpuno besplatno. Tako da je i ovaj overview besplatan. P.P.S. Naravno, nisam zapravo znanstvenik lmao.

Krešimir ĆorlukaSep 14, 2026

During the summer, I decided to solve SEO/AEO/GEO forever by doing SCIENCE. 🔬 I collected over 30,000 documents from the open web. I was looking for case studies, split tests and academic work that would tell me what REALLY works and what doesn't. I then created a book which is free, and I'll explain why at the end. 👇 Out of those 30k documents, around 9k were about an actual tactic rather than a benchmark stat (AIO eating up CTR and similar). And only 1,391 of them reported an outcome someone had actually measured. Those gave me 7,329 findings, which grouped into 1,105 evidence clusters, and 182 of those earned an entry in the book. Sadly, a lot of the tactics don't have controlled tests behind them. Of the 769 tactics I scored, 602 don't have a single one. 138 have exactly one, and only 6 have three or more. The top tier, the "REALLY LEGIT" one that needs two or more independent controlled tests, has 8 entries in it. But shoutout to SearchPilot, they do this properly. Of the 118 controlled measurements in the whole review, 65 are theirs. ✌ I'm not calling out anyone here. Controlled tests can be hard to manage with clients, sometimes they're expensive, and, probably, a lot of agencies aren't really up for publishing the exact methodology they used. Publishing something like "we boosted a client's revenue by 300% with anchor text optimization" is more than good enough. You don't have to tell your competitors exactly how you did it or how you tested it. So, I can't really say I "solved SEO", because there simply aren't enough controlled tests out there to see what REALLY works. But still, everything I've gathered is worth taking a look at, and please do NOT think you can just hit "apply all" on these tactics. You need an expert to check in on that. If you want the ebook just comment "ENGAGEMENT BAIT" in the comments and I'll send you a link. Just kidding, the link is in the first comment. Got to get a little cheeky with this. 😅 P.S. Some people tried to convince me not to give this away for free, because the methodology is mine and I put a lot of money and more than two months of work into it. There was certainly an argument for charging for it. However, I don't really think it would be fair to all of the great agencies and companies who shared all of this amazing information for free. So, this overview is free. P.P.S. Also, I'm not really a scientist lmao.

Krešimir ĆorlukaSep 16, 2026

Od svih testova iz velike SEO meta studije, ovo su mi među najzanimljivijima. I izgleda da sam u pravu za disavow. 😁 Mislim, kako bi se reklo, Ovisy. Disavow funkcionira ako zaista postoji ručni penal na webu. Ako ne, onda nema konkluzivnog efekta jer Google dosta dobro ignorira SPAM linkove i izuzima ih od efekta na rangiranje (generalno). Slično je i za AI sadržaj da nema negativnog efekta ako se koristi za stvarno dodavanje konteksta, a ne samo masivno dodavanje sadržaja bez svrhe i smisla. I ofc, ako nađete koju grešku (ipak sam čovjek), slobodno javite. Cijela studija u prvom komentaru! 👇