← Back to blog

Your Tweet Has About 6 Hours to Matter — Here's What the Algorithm's Time Decay Actually Means

Strategy

There's a number buried in how X ranks content that explains a lot of confusing creator behavior: why people repost old tweets, why "best time to post" content never stops trending, and why a good tweet posted at the wrong hour can underperform a mediocre one posted at the right hour. That number is six hours — and it's apparently how long it takes for a post to lose half its potential visibility score.

What time decay actually means

Multiple breakdowns of X's ranking system describe a steep "time decay" function: a post's algorithmic visibility score drops by roughly half every six hours after it's published. By the 24-hour mark, most analyses agree the post has essentially exhausted its organic distribution window — what you see after that point is mostly residual, not active algorithmic push.

This isn't unique to X — most ranking algorithms favor freshness to some degree — but the steepness here is notable. A post that's "doing fine" at hour 2 and only mediocre by hour 8 isn't failing; it's just running out the natural clock every post is on.

The exception that makes this more interesting

Here's the part that actually matters for strategy: this decay isn't a hard ceiling. If a post gets a late burst of engagement — say, someone with a large following replies or shares it twelve hours in — several sources describe the algorithm "re-expanding" distribution in response. The clock doesn't just keep ticking down regardless of what happens; a strong enough signal can restart part of the climb.

This is also reportedly why threads and posts with ongoing replies behave a little differently than single static posts — every new reply is itself a small new engagement event, which can extend the active window a bit longer than a post that gets all its engagement in one immediate burst and then goes quiet.

What this means for when you post

A few practical implications, taken together:

Posting time matters more than people give it credit for, but not because of some magic "best hour" — it matters because the first 1-2 hours largely determine how far that initial decay curve lets you climb before gravity sets in. Posting when your specific audience is actually online gives you a better starting position on a clock that's already running.

Reposting the exact same content immediately after it underperforms isn't a great move — several analyses note this can actually trigger spam-detection behavior rather than giving you a second chance. If something flops, the better move is usually to let it go and try a different angle later, not to retry the identical post.

Engaging with your own replies early matters more than it seems. If a two-way reply chain genuinely carries outsized algorithmic weight (as multiple breakdowns suggest), then showing up in your own comments in that first hour isn't just good etiquette — it may be actively extending the life of the post itself.

The honest caveat

As with most of these algorithm breakdowns, the precise decay curve and exact re-expansion thresholds aren't something X has officially confirmed down to the exact numbers — these are informed estimates from people who've studied the open-sourced code and observed real account behavior over time. The six-hour figure shows up often enough across independent sources that it's worth taking seriously as a general shape, even if the exact curve isn't guaranteed to be precisely that.

The simple version

Your post has a short, real window where it's actively being pushed — roughly the first half-day, with the first couple of hours mattering most. After that, you're mostly coasting on momentum it already built. Knowing that changes the question from "is this a good tweet" to "did I give this tweet a fair shot in the hours that actually counted."

---

*Not sure what to post and when momentum is slipping? TweetGem helps you turn an idea into a tweet fast, so you're not losing your best posting window while you stare at a blank box.*