How X's Recommendation Engine Actually Works Right Now — And What Smart Creators Are Doing About It
If you've noticed your impressions tanking or your engagement behaving weirdly, you're not imagining things. The recommendation engine behind Twitter/X has gone through some real structural changes since 2024 kicked off — and the playbook that worked in 2022 or even early 2023 is basically outdated at this point.
At Twiplog, we spend a lot of time digging into the numbers behind social media behavior, so let's cut through the noise and talk about what's actually happening under the hood.
What Actually Changed in 2024
The biggest shift has been a move away from pure follower-based reach toward what X calls "interest graph" signals. In plain terms: who follows you matters less than how people interact with your content — and specifically, whether people who don't follow you are engaging with it.
X's open-sourced recommendation algorithm (yes, they actually released parts of it) confirmed that the system uses a tiered scoring model. Replies carry significantly more weight than likes. Bookmarks — a signal that users often overlook — have become a surprisingly strong quality indicator. Reposts (formerly retweets) still matter, but they've been downweighted slightly compared to active engagement like quote posts and threaded replies.
There's also been a notable push toward rewarding "heavy Twitter users" — accounts that post consistently and stay on the platform. If you're the type who fires off one tweet a week and disappears, the algorithm is basically treating you like a tourist.
The "For You" Feed Is a Different Beast Now
The For You page has become X's primary battleground for organic reach, and it operates very differently from the Following feed. Content served there is heavily influenced by:
- Engagement velocity in the first 30 minutes — posts that pick up fast replies and bookmarks early get pushed harder
- Semantic relevance — X is better at categorizing topic clusters than it used to be, which means niche content actually has a fighting chance if it's consistent
- Account health signals — things like posting frequency, reply behavior, and whether your account has been flagged for low-quality interactions
One thing that's often overlooked: external links have been quietly penalized in terms of reach. Posts that push users off the platform — to a newsletter, a YouTube video, a blog post — tend to get less distribution. That's a real tension for creators trying to drive traffic somewhere.
Case Study: The Niche Finance Account That Tripled Impressions
A personal finance creator based in Austin, Texas — running an account in the 8,000-follower range — noticed their impressions dropped about 40% in Q1 2024. After auditing their posting habits through a third-party analytics dashboard, they found that about 60% of their posts included external links.
They ran a 30-day test: no external links in the main tweet body. Instead, they moved links to the first reply. Impressions recovered by the third week and ultimately climbed 3x above their pre-dip baseline. Engagement rate improved from around 1.8% to 4.1%.
The lesson isn't that links are dead — it's that where you put them changes your distribution math entirely.
A Simple Framework for Testing What Works in Your Niche
Here's a lightweight testing structure you can actually stick to:
Week 1-2: Establish your baseline. Post at your normal cadence and log your average impressions, replies, bookmarks, and profile visits per post. Don't change anything yet — you need clean data.
Week 3-4: Isolate one variable. Pick one thing to change: post timing, format (thread vs. single tweet), media inclusion, or link placement. Keep everything else constant.
Week 5-6: Measure and document. Compare against your baseline. Did engagement velocity in the first 30 minutes improve? Did you see more For You reach? Log the delta.
Repeat the cycle with a new variable. Over a few months, you'll have a niche-specific playbook that's built on your actual data — not generic advice.
What to Actually Optimize For
Stop chasing likes. Seriously. Based on how the algorithm is currently weighted, here's a rough priority order for what to focus on:
- Replies — especially back-and-forth conversations that keep users on the platform
- Bookmarks — a strong signal that your content has lasting value
- Quote posts — drive new eyeballs and signal that your content is discussion-worthy
- Reposts — still valuable, but less so than active engagement
- Likes — they register, but they're the weakest signal in the mix
Building content that invites a reply — through a question, a hot take, a counterintuitive stat — is one of the most reliable ways to improve your distribution score over time.
The Long Game
Here's the honest truth: there's no single hack that unlocks the algorithm. What the data consistently shows is that accounts growing steadily in 2024 are doing a few boring things really well — posting consistently, engaging genuinely in replies, and tracking their metrics closely enough to notice when something shifts.
That last part is where a lot of creators fall short. If you're not logging your performance data regularly, you're essentially flying blind. The algorithm will keep changing. Your job is to notice when it does and adjust before your reach craters.
That's the whole game.
