How to A/B test Instagram reels with trial reels

Trial reels are reels Instagram shows to people who don't follow you, before (or instead of) your followers. That's what makes them work for A/B testing: you post a few versions as trials, see which one strangers actually watch, and only the winner reaches your followers.

Instagram launched them in December 2024. A few things about how they behave matter for testing:

• Your followers won't see a trial in their Feed or Reels tab.
• It's not on your grid or your profile's Reels tab, unless you share it with everyone later.
• Insights show up within 24 hours: views, likes, comments and shares.
• A few followers can still see it, if someone sends it to them in a DM, or on a page that groups reels by the same audio, location or filter.

There's no "A/B test" button on Instagram. You run the test yourself, one trial per version. That's fine. It just means the method is on you.

You need a professional account, creator or business. Instagram's help page adds that some creators without one may get access as part of ongoing testing.

You'll see follower numbers all over the place: 1,000 on one blog, 200 on another. Instagram's own help page doesn't name one. If the Trial toggle isn't on your publish screen, check you're on a professional account first, then make sure the app is up to date.

Make the reel as usual and flip one switch before you share it.

1. Create a reel and follow the steps until it's time to share.
2. Tap the toggle to turn on Trial.
3. Tap Share.
4. Within 24 hours, open the reel and tap View trial insights to see how it's doing.
5. When you've picked a winner, tap Share to everyone on that one.

There's also a setting to let Instagram share a trial for you. Under the Trials toggle, tap How it works, then turn on Share to everyone automatically. Instagram shares the reel if it gets a lot of views in its first 72 hours. That setting sticks for every trial you post after, so turn it off while you're testing. You want to pick the winner, not have Instagram share whichever version took off first.

Test the hook first. The first second or two decides whether someone keeps watching or flicks past, and it's the cheapest thing to change.

Change one thing per version. If version B has a new hook and a new song, and it wins, you've learned nothing you can use next time.

Good first tests, roughly in order:

1. The hook: a question, a bold claim, or starting mid-action.
2. The on-screen text in the first frame.
3. The length: the same reel trimmed tighter.
4. The sound: a trending track or your own voice.
5. The cover and caption, last. They matter less for strangers scrolling Reels.

One warning. Instagram says a trial may get limited reach if it spots content you've already shared. So each version needs to be a real edit, not the same file uploaded twice. Change something people can see or hear in the opening seconds.

Wait 72 hours, and wait until every version has a fair number of views. Insights arrive within a day, but a reel that's ahead at hour six is often behind by day three.

Seventy-two hours isn't our number. It's the window Instagram uses to decide whether to auto-share a trial, so it's a sensible length for your own test too.

Post every version the same day, close together. If version A goes up Monday morning and version B on Wednesday night, you're testing the day as much as the hook.

As a rule of thumb, we don't call a test until each version has about 1,000 views. That's our habit, not an Instagram rule. Below that, one person sharing a reel to a group chat can swing the whole result.

Pick on whether people watched, not whether they liked it. A like is cheap. Watching is what tells you the hook worked.

If your insights don't show a skip rate, go on watch time alone. Compare each version against your usual reel, not just against each other. If all three trials did worse than your normal reels, the winner is the least-bad version, and the next round should try something bigger.

That's the whole method, done by hand. If doing it every week sounds like a second job, that's the part TopPost does for you: it makes the versions, posts them as trials and reads the results.

Instagram says no. Its help page says trial reels "are treated separately" from your other reels in its ranking systems, so they "shouldn't affect the performance" of your regular reels.

Two honest caveats:

• Trials can take longer to get views, because Instagram is hunting for non-followers who'll like them. A slow start isn't a punishment.
• Reposting the same content gets limited reach. That's the real way to hurt a test: lazy duplicates.

People on Reddit swear their reach dipped after posting trials. Maybe. We'd trust Instagram's own statement over a hunch from one account, but watch your regular reels' numbers for a few weeks and decide for yourself.

For Instagram, trial reels. Facebook has a proper A/B tool, but it only covers Facebook.

If you post the same reel to both, test on Instagram with trials and let the winner go to Facebook afterwards.

Most failed tests come down to one of these.

• Uploading the same file twice. Instagram can limit reach on repeated content. Make each version a real edit.
• Changing three things at once. You'll get a winner but no lesson.
• Posting days apart. Timing becomes the variable.
• Calling it at hour six. Wait the 72 hours.
• Leaving auto-share on. Instagram may share a version before your test is done.
• Testing a dud. If the idea is weak, no hook saves it. Test hooks on reels you'd be happy to post anyway.

TopPost turns the method above into a loop. It's built for short-form video from your own footage: it edits your clips, and its agent makes the test versions, each with one change. Then it posts them to Instagram as trial reels through Instagram's official API.

After 72 hours it reads the round. Each version is scored against your account's usual numbers, on fewer early skips and longer watch time, with shares and saves breaking ties. It waits until every version has enough views before it calls anything.

Then the winner is yours to share. Instagram's API can't share a chosen trial to your followers, so that last step is one tap in the Instagram app. From there, TopPost schedules the winner and can turn the comments into leads with auto DMs.

What it isn't: an AI video generator. You bring real footage, and it's vertical 9:16 reels and shorts only. Trial reels are an Instagram feature, so the testing only works there.

Pick your next reel and make two versions with different hooks. Post both as trials the same day, turn off auto-share, and come back in 72 hours. That's your first test.

If you'd rather not run it by hand every week, try TopPost and let the agent make the versions and read the results for you.