First 15 minutes decide the next 48 hours.
Engagement velocity in the opening window is the single largest variable in whether a post escapes your follower graph. Timing matters; velocity matters more.
First 5 minutes
The Heavy Ranker has not yet collected enough engagement data to differentiate your post from baseline; it relies on Thunder + author embedding alone. This is the window where in-network readers — your most reliable repliers — should engage if you want the post to escape your follower graph.
Minute 5–15 · the velocity decision
The single most consequential interval. The ranker re-scores your post against accumulating engagement and decides whether to start serving it to Phoenix candidates. A self-reply with new content, a strong early reply from an in-cluster account, or a quote-tweet from an embedding-adjacent account in this window can 10× the eventual reach. Late engagement (over 30 minutes) is still helpful for the post's lifespan but does not change the ceiling.
Hour 1 · the trajectory lock-in
By minute 60 the post's distribution trajectory is essentially locked in. If you have not crossed the threshold for Phoenix out-of-network distribution by the end of hour 1, you are unlikely to. Posts that go viral hours or days later are statistically rare (~3% of viral tweets); the vast majority of viral posts were already on a viral trajectory by hour 1.
24–48 hours · the second wave
If a post does sustain into a second day, it is because Phoenix's two-tower retrieval has identified strong embedding similarity between your post and content that previously performed well in adjacent clusters. The second wave is a re-surfacing event, not a continuation. Plan content for the second day only after you have confirmed the first-day trajectory.