Logo

X/Twitter Algorithm: Why Replies Are 150x More Valuable Than Likes

In April 2023, Twitter open-sourced its recommendation algorithm. The data revealed something most marketers still don't fully appreciate: a reply engaged by the author carries 75.0 weight vs a like's 0.5 weight -- a 150x difference. This isn't opinion. It's code. Understanding these weights fundamentally changes how you should approach X/Twitter growth.

What Is an AI Social Media Manager?

When Twitter open-sourced its algorithm in April 2023, it revealed the exact scoring weights used to rank tweets in the For You feed:

  • Reply engaged by author: 75.0 -- when someone replies to your tweet and you engage back, this is the strongest positive signal
  • Standalone reply: 13.5 -- even without the author engaging back, a reply is 27x more valuable than a like
  • Retweet: 1.0 -- only 2x the weight of a like
  • Like: 0.5 -- the weakest positive engagement signal
  • Negative feedback: -74.0 -- hiding or muting content heavily penalizes it
  • Report: -369.0 -- the strongest negative signal

Important caveat: These are the published 2023 weights. In January 2026, X released a new Grok-powered recommendation algorithm, but has not published the updated weights. The directional principle -- replies significantly outweigh likes -- almost certainly still holds, as it aligns with X's stated goal of promoting conversation.

Why Most Tools Fail

Most X/Twitter growth tools are built around the wrong engagement signals:

  • Like-based strategies are nearly worthless -- at 0.5 weight, mass-liking barely registers in the algorithm
  • Scheduling tools ignore engagement entirely -- they help you post but do nothing to generate the replies that actually boost reach
  • Manual engagement doesn't scale -- you know replies matter, but manually writing thoughtful replies to 30-50 posts daily takes 1-2 hours
  • Generic engagement tools automate likes -- the lowest-value signal, essentially wasting API calls
  • Timing matters enormously -- the first 30-60 minutes after a tweet is posted are critical for algorithmic ranking. Manual engagement can't consistently hit this window

Additional algorithm factors most tools ignore: - X Premium gives a 2-4x algorithmic boost -- paid subscribers' content gets preferential treatment - External links reduce reach 30-50% -- linking out in tweets is algorithmically penalized

The Amplifresh Approach

Amplifresh is built around the algorithm's most powerful signal: replies.

  • Automated replies on others' posts -- Amplifresh focuses on generating and sending replies (13.5-75.0 weight) rather than likes (0.5 weight)
  • Fast response timing -- automated engagement hits the critical 30-60 minute window after posting, when algorithmic impact is highest
  • AI-powered quality -- replies are contextual, relevant, and trained on your brand voice. Quality replies are more likely to receive author engagement, triggering the 75.0 weight multiplier
  • Targets high-value conversations -- engage with posts from accounts in your target market where your replies get maximum visibility
  • Volume without sacrifice -- scale to 30-50 quality replies daily without spending hours in your feed

The math is simple: 50 automated replies per day at 13.5 weight each = 675 total weight. Compare that to 500 likes at 0.5 weight = 250 total weight. Fewer actions, nearly 3x the algorithmic impact. Get started free with Amplifresh.

Who This Is For (And Not For)

Ideal For

  • B2B marketers who want to understand X's algorithm for strategic advantage
  • Founders who know replies matter but lack time for manual engagement
  • Growth teams optimizing their X/Twitter strategy with data
  • Anyone tired of mass-liking strategies that produce no results

Not For

  • Those looking for a quick hack to game the algorithm
  • Brands unwilling to invest in genuine, quality replies
  • Users who want to grow through likes and retweets alone

Frequently Asked Questions

According to Twitter's open-sourced algorithm (2023), a reply engaged by the author has a weight of 75.0 vs a like at 0.5 -- that's 150x more valuable. Even a standalone reply without author engagement has 13.5 weight, which is 27x more than a like.

Curious how Amplifresh could work for your specific situation?