Which Instagram reels are worth learning from

You save a reel because it did huge numbers, you make your own version, and it does nothing. That's not bad luck, it's usually that you learned from the wrong reel.

Big numbers on a big account mean nothing

A reel with two million views from an account with three million followers is a normal day at the office. It tells you the account is big. It doesn't tell you the idea was good.

What you want is a reel that beat its own account's usual numbers by a wide margin. That gap is the actual signal, because the audience size was the same and something about that one reel made people behave differently.

Look at accounts your size, or a bit bigger

The people one or two steps ahead of you are the useful ones. Their audience is close enough to yours that what works for them is likely to work for you. A giant account is playing a different game with different physics.

Read the first two seconds, not the whole thing

Most of what makes a reel work happens before anyone decides to stay. The opening frame, the first words, whether something is already moving. Watch the first two seconds of ten strong reels back to back and the pattern gets obvious fast.

Then notice what you can actually do. If every one of them opens on a face talking directly to camera and you hate being on camera, that's useful information about which reels to learn from and which to skip.

Copy the shape, never the words

The transferable part is the structure. A reel that opens with a problem, shows one specific example, and ends without a summary is a shape you can put your own topic into.

Copying the words gets you a worse version of something that already exists. Copying the shape gets you your own thing that happens to work.

Make this a habit, not a project

The people who get good at this are checking a short list of standout reels every week, for ten minutes, rather than doing a big research session once and forgetting it. Kadenza keeps that shortlist for you so the ten minutes is actually ten minutes.