Content skills
Every niche already has posts that beat their account's usual by a wide margin. Learning why is the quickest way to improve. Play a few rounds below, then read why the winners won.
Try it here
Two real posts from the same kind of business, likes hidden. Pick the one that did better than its own account usually does. After each pick you see the real numbers and a short reason for each post.
Pick a kind of business. You get real posts from accounts in that niche, likes hidden.
What counts as a fair round. Reels only play reels, and photos and carousels play each other. The winner has at least 3x its account's usual likes and the loser a third or less (or 4.5x against half). Pinned posts, giveaways, paid partnerships, collabs and anything under 48 hours old are left out, and the winner's comments have to be above its usual too, so a boosted post can't win on likes alone.
Expect to miss some. In our early rounds, players picked right about 6 times in 10 (a small sample). That's what makes the reasons worth reading.
The short answer
The game above does this for you. By hand, it goes like this: collect 20 to 30 posts from accounts that sell to your audience. Compare each post to its own account's usual likes, not to anyone else's, and keep the ones that beat it by 3x or more. Put them beside the same accounts' flops. Then name what each winner did in three parts: what made people stop (the hook), what kept them watching, and whether the ending paid off the opening. Keep that mechanism, drop the surface, and make something only you could make. Do it with a guess first, before you look at the numbers, or you won't learn much.
After you pick
Each reason covers the same three things you'd write down by hand.
What gave a stranger a reason to stop: the topic, the opening words, and whether it looked like the posts that niche already likes.
What kept them watching: a list with the number said up front, steps toward a result, or a story with an open question.
Whether the ending paid off exactly what the opening promised. A hook that promises more than the post delivers loses the viewer for good.
The losing post gets one too, and it's often the more useful of the two. “Good lighting, but the point only arrives at second twelve” teaches more than a winner's highlight reel.
Where people go wrong
When you open a post you already know did well, every detail looks like a reason. The text is punchy, the lighting is nice, the person is likeable. You'll find a story for any post, and the same story would fit a post that flopped.
The fix is to commit to a guess first. Look at two posts without the numbers, say which beat its account's usual, and only then check. Being wrong is the useful part. It shows you which of your instincts about hooks and formats don't hold up in your niche, and you can't see that when you only study the winners.
That's hard to set up by hand, because you need pairs where the answer is hidden and the comparison is fair. It's why the game is built the way it is.
Your own niche
The sample decks cover a few kinds of business. What helps you most is the posts from the accounts your own audience actually follows. If you sign up and add your Instagram account, SocialToolsAI finds those competitors and builds the same rounds from their real posts. The reasons for each round come with it, depending on your plan, and your rank moves as your eye improves.
You can get the same result without us. The manual method further down works on your competitors too, with a spreadsheet and an afternoon. The game removes the collecting, the arithmetic and the blank page where the reasons go. No Instagram login is needed.
Prefer to do it yourself?
You don't need a tool for this. It takes an afternoon the first time, and it works. The game just does the collecting and the arithmetic for you.
Pick five to ten accounts that sell to the same people you do. Take their top posts from the last few months. Fewer than 20 and you're looking at luck, not patterns. Don't only take the viral ones.
Raw likes lie. A 4,000-like post from an account that usually gets 3,500 did nothing special. A 900-like post from an account that usually gets 150 did something. Write down each account's usual (the middle of its last dozen posts) and keep only the posts that beat it by 3x or more.
Find posts from the same account that landed well under their usual, and set them beside the hits. Same person, same audience, same month. Whatever differs is where the lesson sits.
The hook: what made a stranger stop (the topic, the first line, the way it was filmed). The retain: what kept them watching (a list, steps, or a story they wanted the end of). The reward: whether the post paid off exactly what the hook promised.
Copying a winner's words, jokes or props gets you a weaker copy. Keep the mechanism: the kind of topic, the shape of the opening, how it holds attention. Then make a post that only you could make with it.
Questions
Twenty to thirty from a handful of accounts is a workable minimum. Below that a single lucky post can look like a rule. The point is to see the same hook or format win more than once, in more than one account.
Because audience size distorts both. A post with 5,000 likes can be a flop for an account that usually gets 20,000, and a post with 600 can be a hit for one that usually gets 100. Comparing a post to its own account's usual strips out the size of the audience and leaves the part you can copy: what the post did differently.
Test it. In our early rounds, players picked the better post about six times in ten (a small sample). That's better than a coin flip and a lot worse than most people expect. If you score higher, good: your rank climbs.
Yes. The five steps on this page work with a spreadsheet and an afternoon. The slow parts are collecting posts, working out each account's usual, and writing down why a post worked. The game takes care of the first two and gives you its reading of the third after you've made your own guess.
The real likes and the account's usual for both posts, and a short reason for each in plain words: what the hook was, what held the viewer, and whether the payoff matched. The reasons are written by an AI reading the posts, so treat them as a second opinion to argue with, not a verdict.
No. Playing here needs no account and no Instagram login. Signing up to a SocialToolsAI plan lets you play with posts from your own competitors instead of the sample niches.
Not if you keep the mechanism and drop the surface. A hook that works because it opens on the result and then rewinds can be used for a dentist, a florist or a bookkeeper, and the posts will look nothing alike.