Six months ago, you could game the X algorithm.
The entire codebase was open-sourced on GitHub. You could read the exact scoring rules, reverse-engineer what the algorithm rewarded, and build content designed to hit those thresholds.
Likes weighted heavier than retweets? Post accordingly.
Certain phrases getting deboosted? Cut them out.
It was pure, predictable “if-then” logic
That's over.
A few months ago, X switched to what they're calling the Grok AI algorithm. Every single piece of content you post now gets read by Grok before it goes anywhere. It's not checking a checklist. It's evaluating whether your content is actually valuable and relevant to the people who should see it.
So if your reach has dropped, it's not a shadow ban. Grok just decided your content isn't worth pushing. That's blunt, but it's the reality.
Here's what changed and what you can do about it.
Custom Timelines Change Everything
X just rolled out a feature where users can pin up to 75 topics on their For You page. Someone interested in AI, business, or real estate can now build a timeline that shows them only that. No politics, no noise, no content they didn't ask for.
What this means for you:
Grok is now matching your content to specific topic feeds. If you consistently post about one topic, Grok learns what you are and starts placing you in front of the people filtering for it.
The broad "something for everyone" strategy doesn't work anymore. Pick your topic, stay in your lane, and let Grok do the matching.
The Compound Effect
The moment you go viral on X, your next one or two posts immediately get shown to everyone who just engaged with that viral piece. Grok is essentially probing the audience: you liked this, you'll probably like them again.
So what we do at The Birdhouse: the second a client goes viral, the next scheduled post is a CTA, a lead magnet, or a conversion/BOF post. Not a personal update. Not filler. Something designed to capture the people riding the wave.
Back-to-back virality is real. We see it constantly.
Same logic applies to replies. If you reply to Elon or a big account and pull 100K views, your next post gets a boost too. Use it intentionally.
The Format That's Crushing Right Now
Quote tweets on articles.
Find a high-value article in your niche, drop a quote tweet on it with your take, and you tap into two feeds at once. Your audience sees it, and anyone browsing that article's "view quotes" page sees it too.
The reason it compounds: once someone engages with a quote tweet on an article, the algorithm floods their feed with similar ones. It's a chain reaction. One interaction and they're seeing five more back to back.
We've been running this format for clients. The views-to-inbound-leads ratio is the best we've seen on any format right now.
What Actually Kills Your Reach
Three things Grok punishes:
Engagement spam. If you're posting "good morning" fifty times a day or copy-pasting the same reply across accounts, Grok catches it. Head of Product Nikita literally screen-records people's profiles when they complain about shadow bans. It's always this.
Conflict and outrage bait. Grok is deprioritizing sensationalism. Leaning into breaking news, war, politics, anything that feels like engagement farming? It's getting pushed down.
"BREAKING 🚨" in all caps. X confirmed this reduces your creator payouts by up to 60%. If it's cutting your payouts, it's probably cutting your reach.
The algorithm isn't the problem. It never was.
It's just another word for audience. Grok is doing what every platform eventually does: optimizing for content people actually want to see. So if you're not getting pushed, the content isn't good enough yet, or it's not specific enough.
Both are fixable.
Hope this helps.
Cheers.
Marcos
P.S. I put together a 14-minute YouTube video walking through every piece of this. Link here if you want the full breakdown:
