When I started ghostwriting for Gianluca and Jason in September 2025, the playbook was simple: post regularly, write in their voice, chase reactions. It worked for four months.

Then, toward the end of 2025, LinkedIn quietly swapped its entire ranking system for a new AI model, 360Brew. Overnight, the rules I'd been optimizing for stopped applying. Fast reactions in the first hour? No longer the signal that mattered. The platform got noticeably better at sniffing out anything that read like AI-generated filler (Seems like AI slop button? Mhmm) and punished it. I had to unlearn almost everything I thought I knew and figure out, in real time, what "performing well" even meant anymore.

Platform-wide, views dropped roughly 50% and engagement dropped roughly 25% after the shift. Everyone's numbers fell. So when I tell you Gianluca and Jason's accounts didn't dip through this, that's not a small thing. It means the content held its ground while the algorithm was actively serving less content to everyone, not that it coasted through unaffected.

The real lesson wasn't a single tactic, it was learning to distrust the metric I used to trust most. Reactions used to be the whole scoreboard. Now the signals that actually predict whether a post is doing its job are saves, substantive comments, and reposts - the things that mean someone stopped, thought about it, and did something with it.

That's the rulebook I rebuilt from, and it's the one I still use.