Aubrium Blog
You are a cold item: what Instagram's own documents say about a new post
Everyone repeats that Instagram tests a new post on 500 people. That number is in no Meta document we could find. The 500 Meta does publish counts posts, not people.
Where a clip travels furthest: Shorts vs Reels vs TikTok
The same 30-second clip meets three different engines. One ignores your follower count, one ranks by it, and one publishes almost nothing. A comparison built only from what the platforms themselves publish.
What actually travels
The largest study of its kind found false stories spread faster, farther and deeper than true ones. Not because of bots. Because people respond to novelty and emotion.
The thousand-model machine, explained for humans
Meta's engineers describe over 1,000 models, two-tower networks, and a funnel that narrows billions of posts to hundreds. Here is what that actually means.
The minus sign — what Instagram scores against you
Meta's 2023 engineering post prints the scoring formula. The "see less" term is subtracted. Every system card lists predictions nobody talks about.
Stories is not Feed, and almost nobody researches it
Every guide is about Reels and Feed. Stories runs on a different system with different signals, and it is the one surface built to reach people who already follow you.
Saves and sends beat likes: the quiet signals that move reach
Mosseri said sends matter most. The system cards list them as one guess among many, with no published weights. The loud number everyone chases travels least.
The link tax: does putting a link in a post cost you reach?
The belief that external links tank your Instagram reach is everywhere. We went looking for where Meta documents it. Across four system cards and 44 named signals, it is not there.
The half-life of a post: how long your content actually keeps working
A tweet's working life is measured in minutes. A Pinterest pin's in months. What the platforms actually document about how long a post keeps reaching people, and what they stay silent on.
