I bet this happened to u at least once; someone posted a screenshot of a feature nobody else can see, three people check their apps, nobody has it, and everyone concludes the screenshot is fake or their account is broken, or u forgot to turn on something in settings… Neither is usually true. Instagram tests new features on tiny slices of users, sometimes as small as half a percent, and it uses three different rollout shapes depending on what’s being tested. Adam Mosseri, Instagram’s Head, walked through all three in an August 2026 Q&A.
Shape one: the small percentage
Most small features go out to a fraction of the user base, in the range of half a percent to a percent. The team watches whether people use it at all, and then whether using it changes how they feel about the app or how they behave in it.
Usage alone isn’t the bar. A feature can get tapped constantly and still fail if it doesn’t move anything that matters, which is why “people are using it” is not the same as “it’s working.”
Shape two: the whole country
Some features are worthless in isolation. Anything social, where the value depends on other people having it too, can’t be tested on a scattered half percent because those users would have nobody to use it with. Notes is the example Mosseri gave.
For those, Instagram launches into an entire country and watches how that country adopts it. It’s the only way to test a network feature honestly, since the whole question is whether a connected group takes it up.
Shape three: the creator cohort
Creator tools follow a third path. They go to a small set of creators first, collect feedback, and get iterated on before anything reaches the wider community. That’s a qualitative loop rather than a statistical one, and it’s why creator features often arrive more polished than consumer ones.
The throughline across all three, in Mosseri’s framing, is to “understand what works and what doesn’t, what’s resonating and what is not” before committing. The mechanism differs, the question doesn’t.
Walking it through with a real example
Say Instagram builds a new sticker. It goes to half a percent of users. The team sees strong tap rates but no change in story posting frequency, which suggests the sticker is cannibalizing other stickers rather than creating new posting. That feature probably dies quietly, and nobody outside Meta ever knows it existed.
Now say it’s a group feature. Half a percent of users would each be the only person in their friend group who has it, so tap rates would be meaningless. It goes to one country instead, and the metric becomes whether groups form.
Same company, same goal, completely different test design. The rollout shape tells you what the team thinks the feature is for.
The lever you can actually pull
You can’t opt into tests, and no setting exposes them. What you can do is stop treating a feature you’ve seen once as something you can plan around. A feature in test is not a feature that exists, and it may never ship.
The practical version: when someone on the team spots a new capability in the wild, log it as a watch item rather than a roadmap item, and check whether it’s still there in a month. It’s the same discipline that applies to anything Mosseri says he’ll take to the team, like custom carousel cover photos or search across your own posts, and it keeps your content planbuilt on things that are actually available to your accounts today.
If you want to know which features have genuinely launched, Instagram’s own feature pages are more reliable than a screenshot in a group chat.
FAQ
How does Instagram test new features? Three ways. Small features go to roughly half a percent of users, social features launch in a whole country, and creator tools go to a small creator cohort first.
Why do some people have Instagram features I don’t? Because they’re in a test group. Tests run on small random slices of users, and there’s no way to opt in or request access.
Does Instagram work with specific creators on testing? Yes. Creator features are often released to a small set of creators first so the team can gather feedback and iterate before a wider release.
Will every tested feature launch? No. Testing exists to find out what doesn’t work. Plenty of features are tested and quietly dropped without any announcement.
