I spent two years reverse-engineering Google.
I came out of the Amiga scene — the type who has to understand how a system works from the inside. In the early 2000s, one question wouldn't let go of me: how does Google actually decide what ranks first? I set out to answer it the only way I trusted, by experiment.
I built around fifty sites on a single theme and pushed every one of them into the top 100. Then I tested, again and again. Change one thing on site number ten, wait, and watch: if it slid to twentieth, wrong lead; if it climbed to seventh, I was onto something. Nothing genius about it — a spreadsheet, patience, and two years of method. By the end, I could put more or less whatever I wanted on the first page. After thousands of hours of work, the reverse-engineering had worked. I had found 87 precise criteria to rank on Google's first page.
I built my SEO company on that. I never went looking for clients; ranking in the Top 5 for thousands of keywords, they came to me — and many are still here a decade later, which is exceptional in the SEO business. The heart of it was white-label travel comparison engines, in the Kelkoo mold, in the early days of online travel: first to third on “voyage Maroc”, “achat billet d'avion”, and thousands of other queries, with sites running in French, German, English, and Spanish. One deal-of-the-day site of mine reached a million unique visitors a month in 2005.
Back then, first meant first — no ads stacked above the natural result. The surface keeps changing; the discipline doesn't. The same method that read Google's ranking in 2004 is the one I now turn on AI answer engines.
— Benoist Rousseau