How does a good "study of what is working" on social media looks like?

If this post doesn't follow the rules, please report it to the mods.

I am a bot, and this action was performed automatically. Please contact the moderators of this subreddit if you have any questions or concerns.

I would start by defining the decision you want the study to help you make. For example, do you want better hooks, more saves, more profile visits, or more comments? Otherwise it is easy to collect interesting examples without learning what to repeat.\n\nFor each post, I would record the audience or topic, the promise in the first line or second of video, the format, the length, the first visual change, and the call to action if there is one. Then compare each post with your own recent baseline rather than chasing the biggest account.\n\nAfter a few weeks, group the results by pattern instead of treating every post as unique. A hook may win attention while the post fails to create profile visits, which is a different lesson from a hook that brings fewer views but much stronger saves or conversations.

I still use the same basic structure, but I get faster at deciding what to measure. Once you have a few weeks of examples, the review becomes less about documenting every detail and more about checking a short list of signals that match your goal, then digging deeper only when something stands out.

the easiest way is to stop writing down “good hook.” that’s too vague to reuse.

for each post, log five things: the topic, the promise in the first line, the visual in the first two seconds, how the idea is proved, and what people are saying in the comments.

then compare it with that creator’s normal posts, not just their biggest hit. a reel with 50k views means something different on a 5k-view account than on a 500k-view account.

after 15 to 20 posts, look for repeats. maybe list hooks beat questions, before-and-after visuals hold better, or examples get more saves. turn only those repeats into a small test for your own page.

also write down why you think each post worked before checking the comments. that stops the notes becoming a random swipe file and forces you to build an actual opinion.

One thing that made this much less overwhelming for me: you can't see anyone else's retention, so studying their hooks on their own is guessing at a layer you have no data on.

What you can see from outside is the spread. Pull an account's last 20 posts, find their median view count, and only study the three or four that beat it by a wide margin. The rest is their baseline and tells you nothing.

Then the comparison is narrow enough to be useful: what did those three do in the first second that the other seventeen didn't. Usually it isn't the words, it's what's on screen while the words happen.

Doing it on your own posts first is better still, because there you have the retention graph and you can check whether people stayed or just showed up.

A useful way to study this is to turn every observation into a testable hypothesis, not just a swipe file.

I’d tag 20–30 posts with only six things: audience problem, hook type, first visual, content structure, proof used, and CTA. Then add one sentence: “This probably worked because ___.” Look for a pattern that appears at least 3 times before treating it as meaningful.

The important next step is to test one variable at a time on your own account. For example, keep the topic and body similar but try three hook styles: direct promise, contrarian statement, and open loop. Compare retention at 3 seconds plus saves/shares, not just total views.

That gives you a repeatable loop: collect examples → tag patterns → write a hypothesis → run 3 versions → keep or discard it. Otherwise studying content can become productive-looking procrastination.