Do Links Actually Hurt Your Reach on X? The Answer Is More Complicated Than You Think

Ask anyone who's posted on X in the last year and you'll get a confident answer: links kill your reach. Post a blog URL directly in your tweet and watch it die in silence; post the same link as the first reply instead, and somehow it does fine. This has become close to gospel in creator circles. But when you actually go looking for where this claim comes from, the story gets messier — and more interesting.

The case for "yes, links are penalized"

There's real data behind this. Multiple analyses tracking distribution data across creator accounts report that posts with links in the main body get somewhere around 30 to 50 percent less initial reach than equivalent link-free posts. Some more recent write-ups go further, claiming that as of early 2026, non-Premium accounts posting links see close to zero median engagement — effectively invisible. The standard workaround, repeated everywhere, is to post your main content as a clean text post, then drop the link in the first reply.

If this is true, it's a meaningful behavioral shift X has engineered: keep people inside the app, don't let them leave for outside content, period.

The case for "actually, no, not exactly"

Here's where it gets genuinely interesting. At least one detailed breakdown of X's actual open-sourced code found something that contradicts the popular narrative: there's no hard-coded link penalty anywhere in the published source. The click-weight value for links is positive, not negative. If a link demotion exists at all, the argument goes, it would have to be something the model learned on its own from training data — not a rule someone explicitly wrote into the system.

Under this reading, links can absolutely perform — and when they don't, the actual problem is usually the post itself (low engagement, no hook, posted at a dead time) rather than the URL specifically. The "links are penalized" belief might be a correlation people noticed and turned into a rule, without it actually being one.

So which is it?

Probably some mix of both, honestly. It's entirely possible for a learned model to behave as if there's a link penalty — depressing reach on link-heavy posts because historically they generate fewer replies and less time-on-platform — without anyone writing "if link: reduce_reach()" anywhere in the code. The practical effect on your reach could be identical either way. The difference matters more for understanding *why* it happens than for what you should actually do about it.

What this means for how you post

Given the uncertainty, the safest approach is still the boring one: if you need to share a link, consider posting the core idea as a standalone post and adding the link as a reply, since this costs you nothing and has wide anecdotal support either way. But don't assume every underperforming link post is doomed because of the URL — if a linked post flops, it's worth asking honestly whether the post itself gave anyone a reason to stop and read it, before blaming the algorithm entirely.

The honest takeaway here isn't "links are banned" or "links are fine." It's that a lot of what gets repeated as algorithm fact online is really pattern-matching dressed up as certainty — and the actual code, where anyone can check, sometimes tells a more nuanced story.

---

*Whether you're writing a link post or a pure text one, TweetGem helps you nail the hook that actually earns the read — the part that matters regardless of which theory turns out to be right.*