Meta tests Facebook and Instagram versions that open on full-screen video

A woman posts a photo of her nephew's birthday cake for the relatives who follow her. In Meta's Instagram test, a relative who opens the app sees Stories across the top and a Reel beneath them. The cake is one swipe away, in the feed of accounts they follow ^[3][4]. The Verge noted that writers, photographers, nonprofits and podcasters had already been pushed into making video to stay visible to the algorithm ^[15]. "It's another to ask your aunt and old college roommate to do the same or get bumped onto a different page, easily forgotten," it wrote ^[15].

Teams that distribute through Meta tend to tell themselves the watch-time figures settle what users want. Meta's recent earnings call cited time spent watching video on Facebook, up 9 percent globally and 10 percent in the US and Canada ^[7]. What users actually do is narrower. Morning Consult found that 81 percent of US respondents had watched an Instagram Reel and 39 percent had posted one ^[12], so the share who posted is less than half the share who watched ^[14]. Watch time shows how long people stayed. The Verge pointed out that it is unclear whether people prefer video over text or images, or simply watch what the platform puts in front of them ^[13].

The pitch is an "experience" headed to "video-heavy countries" first, with a potential US expansion in 2027 that Meta hinted at ^[2]. What is running is a Facebook test for some users in India, as reported by TechCrunch ^[1]. On Instagram, Meta spokesperson Devi Narasimhan told The Verge, the option is live in a handful of countries she did not name ^[10]. Users can switch back to the original layout, Narasimhan said ^[11]. The accounts of sound differ. The Verge described Reels as playing with sound on by default ^[5], while Narasimhan said they unmute once a user swipes or clicks into a video ^[11]. Not all tests become permanent features ^[6].

Meta's ranking work points the same direction as the tests. On the call it cited improvements to Reels ranking and models it says can "identify high-quality new Reels at creation" ^[8]. Threads, which reaches half a billion people a month, barely came up ^[9]. If those models work as described, a new Reel can be ranked for strangers before the poster's followers have seen it.

I think the right move before 2027 is to measure first and hold big production spending until Meta says more about the US. The tradeoff is timing. If the US version arrives on the hinted schedule ^[2], teams that waited will build a video operation under deadline. If the test is dropped, teams that rebuilt early will have paid for a new format on a hint.

A 2x2 sorts the exposure. One axis is where an account's Meta reach comes from today, mostly followers or mostly recommendations. The other is whether the account already publishes short video. Follower reach with no video is the exposed cell: both tests give the first screen to video ^[1][3], and on Instagram the followed feed moves behind a swipe ^[4]. Follower reach with video loses less, because its Reels compete for the new first screen while its photos move down. Recommendation reach with video already depends on Reels ranking, and the tests hand those Reels the opening screen. Recommendation reach without video depends on non-video posts reaching strangers, and both tests put video ahead of them at open. For most teams the first number to pull is the share of reach that comes from followers. On the Instagram layout, roughly that share would sit one swipe down.