Post by Nemanja Zivkovic

Most of my content advice was wrong. I have 1,208 posts to prove it. For three years I told clients the same things. Open with a question. Set the scene. Build carousels, the algorithm loves them. Post native video, it loves that more. Kidding, I never likes carousels, but a lot of people did! TodayI finally checked instead of guessing. I pulled every LinkedIn post from 25 of the best B2B creators alive, the whole of Q2, into one table using Tracker, the new Favikon feature, and dropped it into Claude. 1,208 posts. One question - what actually predicts engagement here? The answers were uncomfortable. Hooks of 1 to 6 words got nearly 5x the engagement of 21+ word openers. Questions were the worst-performing hook type in the entire set. Plain text-on-image beat video by more than four to one. And the creator who posted the least, 14 times all quarter, averaged 5x the engagement per post of the one who posted the most at 116. Posting more didn't buy reach. It just bought more posts. The pattern under all of it - the content winning on a B2B platform barely looks like B2B. Short, human, opinionated, and built by subtraction, not production. I wrote the whole breakdown, every number and named example, in this week's issue of Marketing with a bit of funk. Link below. What's the one piece of content advice you'd want to put to the data?