Dan Petrovic

Dan Petrovic is Managing Director of DEJAN and one of very few practitioners who investigates search systems directly rather than guessing at their behaviour from the results. His method is to work out how a system actually functions, then apply what he finds to technical SEO and large scale optimisation. He has run original research into Google ranking systems, user behaviour as a ranking signal, and how people consume content online.

The technical work covers crawling, indexing, SERP features and Google Discover. Google has twice given him an award for discoveries involving APIs that exposed parts of its internal systems, and he found and reported a core search and ranking security issue that Google then patched. Very few people outside Google have ever found something Google had to fix. His wider research covers contextual link analysis and organic link earning, backed by experiments designed to establish repeatable relationships rather than one off curiosities.

He now applies the same discipline to AI search, looking at how models represent brands and how that representation governs visibility in generative results. He uses mechanistic interpretability to inspect what a model has actually encoded, and model steering experiments to test whether the output can be shifted. It is a genuinely technical line of enquiry, and it joins up with his earlier work on natural language interfaces and Google’s move towards conversational search.

He has presented at SMX, MozCon, SEO Week New York and Semrush Spotlight, and runs advanced workshops and industry training. He has a long association with Griffith University through guest lectures and advisory work. His published research has been recognised internationally, including a place in Moz’s top ten content of the year.