X’s latest experimental rollout on iOS appears to be reshaping the dynamics of how web visits are measured—and critics suggest it may be unintentionally, or perhaps artificially, inflating the perception of online traffic. In recent weeks, several content platforms, including Substack and Bluesky, have detected an unexpected surge of incoming visits that seem less like legitimate user engagement and more like algorithmic artifacts. According to Nick Eubanks, the Vice President of Owned Media at the digital marketing powerhouse Semrush, this unusual rise in what he describes as “fake” views can likely be traced to a newly introduced mechanism within X’s system that preloads webpage content before a user actively decides to click the associated link.
Eubanks, speaking to *The Verge*, characterizes this situation as a textbook example of analytical distortion—an instance where a platform’s product-level experiments unintentionally disrupt the accuracy of its performance metrics. The modification to X’s link-handling behavior changes the way users interact with posts containing external URLs. When someone taps on a link within the platform, X now collapses the post rather than completely replacing it with the browser window. This subtle yet consequential design shift allows individuals to continue engaging with interactive features such as likes, reposts, replies, and bookmarks even as the linked webpage becomes visible beneath. In earlier iterations, however, X’s in-app browser would fully obscure the original post, thereby impeding engagement metrics by preventing users from further interacting with the platform’s interface once they had navigated outward.
Eubanks elaborates that this new preloading process fetches external content in the background before any actual human action occurs, meaning the system effectively simulates a visit long before the user has made the conscious choice to explore that content. This technical shortcut, while perhaps intended to streamline the browsing experience, has a side effect: it significantly inflates website analytics. Metrics such as click-through rates, page views, and engagement counts become exaggerated, misleading advertisers, publishers, and creators into believing their content has attracted heightened interest. In reality, these supposed increases often fail to correspond to genuine human visits or authentic audience behavior.
This phenomenon came as a surprise to some platform operators. Substack’s CEO, Chris Best, for instance, initially celebrated what seemed to be a sudden boom in referral traffic attributed to X. His optimism quickly faded as deeper analysis revealed that the majority of the apparent lift stemmed from automated preloads rather than real readers. Even so, Best later clarified that, after filtering out the artificial data, Substack still experienced a legitimate rise in user activity—though far smaller than the raw metrics first implied. A similar story unfolded at Bluesky, where product manager Paul Frazee reported that X’s updated system had severely disrupted the site’s capacity to measure accurate daily active users, particularly those browsing without logging in. Frazee explained that X’s background loading of links, while designed to accelerate browsing speed, had “ruined” Bluesky’s metrics by generating phantom traffic that appeared authentic at first glance.
Addressing the controversy, X’s Head of Product, Nikita Bier, defended the experiment, asserting that the company’s goal was to resolve a longstanding challenge faced by creators on the platform. According to Bier, posts containing external links historically received lower visibility and engagement because the embedded browser previously covered the original post completely, distracting users from liking, replying, or reposting. The newly implemented behavior, he argued, helps X better evaluate which content deserves amplification by ensuring that engagement signals are not prematurely lost each time someone clicks away to another site.
Although this adjustment may yield short-term benefits within X’s own ecosystem—potentially increasing interaction and time-on-platform—it simultaneously introduces a new layer of complexity for those outside it. Publishers, analysts, and digital marketers now face greater difficulty discerning whether their traffic truly represents human attention or merely automated prefetch activity. As Eubanks warns, the broader digital environment may be entering an era defined by inflated metrics, where user interfaces manipulate statistics through techniques such as content preloading, autoplaying media, and even AI-driven previews or summaries. These mechanisms, while ostensibly improving convenience, also blur the boundaries between legitimate engagement and machine-generated noise. For the sake of maintaining credibility, he concludes, platforms must commit to transparency in how they quantify participation and must clearly differentiate real human actions from algorithmic previews if they hope to preserve trust among creators, advertisers, and audiences alike.
Sourse: https://www.theverge.com/news/816451/x-fake-traffic-preloading-links-bluesky-substack
