Reels: Instagram's discovery format and what it demands in return

Today the distinction is about distribution more than storage: a video published to the feed is also treated as a Reel and enters the video surfaces.

What actually changes is who it is delivered to. A feed post lives mostly off the relationship between two accounts; a Reel is offered by predicted interest in the subject, which is why it reaches people who never heard of you.

That also explains the result asymmetry that startles beginners: two consecutive Reels can land wildly different reach, because each goes through its own initial audience test. In the feed the variation is smaller, because the base is more stable.

According to what the platform states about ranking, the strongest Reels signals are the viewer's recent activity, their history with the poster, information about the reel itself and information about the poster.

Translated into what you control:

• Retention. The strongest behavioural signal, and it does not depend on anyone deciding to act.
• Shares. They carry the Reel to someone through human recommendation.
• Consistent subject. The system needs to know which interest cluster your content belongs to.
• Originality. Reposting and another app's watermark reduce distribution by policy.

The same Reel circulates across different surfaces, each with a different audience.

• The Reels tab. Infinite vertical scroll, mostly from accounts you do not follow.
• Feed. Among the regular posts of people who follow you.
• Explore. The discovery grid, ordered by performance and affinity.
• Profile grid. As a permanent archive, unless you choose not to display it.
• The audio page. Tapping a sound finds every Reel that used it.

Which is why reading a Reel's reach at 24 hours almost always understates: it keeps being distributed for weeks, mainly through the tab and the audio page.

• A hook in the first seconds. The hook decides whether distribution happens, and the retention curve shows whether it worked.
• Readability without sound. A large share of consumption is muted, so the promise has to live in on-screen text.
• Honest length. Stretching lowers retention, which is the main signal.
• No watermark. Video exported from another app with a logo gets reduced delivery.
• Patience when measuring. Judging on day one leads to wrong conclusions.

• I wrote the hook before the rest of the script.
• The video works with the sound off.
• It carries no watermark from another app.
• The length is what the idea needs, with no padding.
• I wait at least a few days before judging performance.