X’s new link experiment on iOS may be artificially inflating traffic. Websites like Substack and Bluesky noticed a sharp increase in “fake” views following the update, something that Nick Eubanks, the VP of owned media at the digital marketing platform Semrush, attributes to a new behavior that preloads content before users click on it.

“What’s happening here is a classic case of metrics distortion caused by product experimentation at the platform layer,” Eubanks tells The Verge. With the new experiment, X will collapse a post when you click on its link, allowing you to interact with the like, repost, reply, and bookmark buttons while viewing the webpage. Previously, when X’s in-app browser opened a link in a post, the loaded page blocked the X post entirely, impacting engagement with the original X content.

“X’s new browser is pre-loading link content in the background, meaning the system fetches the destination page before a human actually taps or views it,” Eubanks says. He adds that this “inflates analytics in a few key ways,” including by increasing the appearance of click rates and tricking advertisers, publishers, and creators into thinking they’re seeing more traffic, when it might not “correlate to real human visits.”

These issues came up for Substack CEO Chris Best, who was initially impressed by the boost in traffic to his site following X’s update before realizing “most of the apparent lift is fake.” However, Best says Substack still saw an increase in traffic “even after correcting for the fake views.”

Paul Frazee, a product manager at Bluesky, also says X’s new preloading system has “ruined” the site’s metrics for measuring logged-out daily active users. “X has started to open links in the background to make them load faster… but it has caused a bunch of other sites to get extra traffic that appears real,” Frazee adds.

X product head Nikita Bier says the new link setup addresses a common complaint from creators, who often find posts with links getting lower reach on the platform. “This is because the web browser covers the post and people forget to Like or Reply,” Bier writes. “So X doesn’t get a clear signal whether the content is any good.”

Though preloading can help boost engagement on X, it may ultimately harm creators and publishers outside X by making it more difficult for them to determine where their traffic is coming from. “We’re entering an era where metrics inflation through interface tricks, preloading, autoplay, and AI summarization will blur the line between user engagement and machine behavior,” Eubanks says. “Platforms that want credibility with creators and advertisers will need to be transparent about how engagement is counted, and separate previews from people.”