X’s decision to hide liker identities (while keeping like counts visible) was widely framed as an engagement play: reduce social friction, make liking feel safer, and watch the “hearts” add up. The platform even nudged users with a product message that implied a direct payoff: liking more would make the “For you” feed better.
But once the dust settled and researchers could compare behavior before and after the June 2024 rollout, a consistent picture emerged: hidden likes on X barely change engagement. The policy may change how people feel about liking, yet the measurable, platform-level outcomes appear surprisingly small.
What “hidden likes” actually changed (and what didn’t)
In June 2024, X shifted Likes from a public signal to a mostly private one. The key detail is mechanical: posts still display the number of likes they receive, but the identities of the people who liked are no longer broadly visible.
According to reporting summarized by the AP, only the author can see the list of who liked their post, while other users cannot. Additionally, users can’t browse another account’s Likes tab the way they could previously, which reduces public traceability of someone’s endorsements.
X Engineering also described the rollout as “making Likes private for everyone,” emphasizing that you can still see your own likes and that counts and metrics still surface in places like notifications. This matters for interpretation: the “engagement metric” (like counts) remains observable, while the “social exposure” (who liked) is what disappeared.
Why X expected engagement to rise
The most straightforward hypothesis was social-cost reduction. If public likes are a reputational signal, then liking controversial, political, or otherwise “risky” content can feel like a public commitment. Remove the public association, and more people should be willing to like.
X engineer Haofei Wang, quoted in coverage of the engineering announcement, argued that “Public likes are incentivizing the wrong behavior… many people feel discouraged from liking…”. Embedded in that rationale is an engagement prediction: fewer social penalties should convert into more likes.
Product messaging reinforced the idea that more likes were desirable and beneficial to the user experience. As reported by the AP, X displayed a prompt telling users, “Liking more posts will make your ‘For you’ feed better,” implying that encouraging likes (and perhaps making them feel safer) would improve both personalization and overall engagement.
What large-scale causal evidence says: no detectable lift
A major reason the June 2024 change attracted attention is that it created a clear platform-level intervention: a before/after moment affecting everyone. That makes it possible to test whether hidden likes produced the engagement bump many expected.
A large-scale causal study released Jan 16, 2026 used a Difference-in-Differences design on 154,122 posts from 1,068 accounts, comparing behavior before and after the June 2024 shift. The line result is blunt: the study found “no detectable” engagement lift attributable to the hiding of liker identities.
Importantly, the analysis specifically examined high “reputational-risk” content, exactly where theory suggests the biggest upside from privacy. Yet it still reported no detectable platform-level increase in likes for that category, and described only a “limited behavioral response” overall. In other words, the predicted effect exists more as a narrative than as a measurable change in aggregate outcomes.
Why surveys show attitude shifts but engagement stays flat
One reason the debate persisted is that user psychology can move even when the metrics don’t. When asked directly, people may say they feel freer to like under a private system, especially when the question foregrounds reputational pressure.
A Jan 16, 2026 within-subject survey experiment of 203 X users found that participants reported “modest increases” in willingness to like when likes are private. That aligns neatly with the platform’s stated rationale: some people do feel discouraged under public likes.
However, the same experiment found no significant change in group-level average likelihood of liking, meaning the intention shift didn’t translate into meaningful engagement change. The behavioral impact remained small, illustrating a common pattern in product changes: what users say they will do often overstates what they actually do at scale.
Reasons hidden likes on X barely change engagement in practice
First, like counts remain public, so a key social signal still exists. Even if identities are hidden, users can still infer popularity from the visible total, and creators still compete for that number. If the count is what drives most behavior (posting strategies, bandwagon effects, “social proof”), hiding identities may not move the needle much.
Second, many likes are not primarily constrained by reputation. People may already like freely within their typical content bubble, or they may avoid liking for reasons unrelated to exposure, habit, feed speed, passive consumption, or preference for other interactions like replies and reposts.
Third, the remaining visibility channels dilute the privacy effect. Authors can still see who liked, and users still receive engagement feedback via notifications and analytics surfaces. That means the change reduces public scrutiny from peers, but doesn’t fully eliminate the sense of being observed, particularly in communities where authors or influential accounts matter more than the general public.
What this means for creators, brands, and analysts
For creators and brands, the practical takeaway is to treat the June 2024 change as a shift in audience comfort, not a guaranteed growth lever. If hidden liker identities reduce friction for a subset of users, it may show up in edge cases or specific niches, but it shouldn’t be assumed to lift overall engagement rates.
For analysts, the policy detail is crucial: like counts are still visible, so time-series engagement dashboards remain interpretable. But identity-level signals (who liked what) are less available to outsiders, changing how competitors, journalists, and third-party tools can audit endorsement networks.
For community strategy, the change can still matter even without a like-count surge. Reduced public traceability may alter how users curate their public persona, how dogpiling dynamics form, or how “receipts” are gathered, effects that are real but don’t necessarily register as a simple increase in likes per post.
The promise of private likes was intuitive: make liking safer, and people will like more. X’s own prompts and engineering rationale leaned into that expectation, framing privacy as both user-friendly and engagement-positive.
Yet the best available evidence points the other way: hidden likes on X barely change engagement. Large-scale causal analysis finds no detectable lift, and survey experiments show that modest intention shifts don’t become meaningful behavioral change. The change may improve how the platform feels, but it appears to do far less for engagement than the original story suggested.
