Sort a large library of high-performing short-form video by what the video is about and you get chaos: skincare next to forklift repair next to sourdough. Sort the same library by how the video is built and the chaos collapses into a short list. The same handful of structures keeps winning in niches that share no audience, no vocabulary, and no product.

Here is the direct answer. What type of content goes viral is the wrong axis. Content does not travel by topic, it travels by shape, and there are roughly five shapes that show up at the top of nearly every niche. The shape is the transferable part. The topic is the part only you can supply, and it is also the part that stops a copied video from working.

That distinction matters because most creators looking for “what goes viral” come back with a topic they have no business making. The better import is a structure.

Why the shape transfers and the topic does not

A shape is a promise-and-payoff structure: what the first second commits to, how the middle sustains it, and what the ending resolves. That structure is doing the work on a viewer who does not know you, does not care about your subject yet, and is deciding in under two seconds whether to keep watching.

None of that is topic-dependent. A viewer’s appetite for a clean before-and-after is the same whether the after is a renovated bathroom or a rewritten cold email. The subject decides who stays past second five. The shape decides whether anyone gets to second five at all.

Topics, meanwhile, are anchored to the audience that already trusts you for them. Borrowing someone else’s topic imports their credibility problem without their credibility. We covered where the ethical and practical line sits in is copying a viral format plagiarism, and the short version is that the recipe travels, the ingredients do not.

The five shapes

These are drawn from how our library sorts analyzed videos into named format classes. There are well over a thousand of those classes, but they cluster: strip away subject matter and most of the high performers reduce to five structural families.

1. The transformation

A visible state change, opened by the end state or a promise of it, resolved inside the runtime. Before and after. Broken and fixed. Cluttered and clear.

This is the highest-ceiling shape in short-form and it is not close. The reason is mechanical: the payoff is legible in a still frame, so it survives muted autoplay, and the open question (“does it actually look like that at the end?”) is strong enough to hold a stranger for twenty seconds.

It travels absurdly widely. Physical transformations, skill transformations, workspace transformations, document transformations. If your work changes the state of anything, you have access to this shape.

2. The build

A process from raw inputs to a tangible finished result. Recipes, repairs, installs, assemblies, a project going from empty file to shipped thing.

The build differs from the transformation in where the value sits. A transformation sells the gap. A build sells the middle, so the promise has to be clear early or the middle reads as chores. In our data the build formats separate winners from losers less on subject and more on whether the finished result is promised up front and then actually delivered on camera. Builds that hide the outcome until the end and take too long getting there are the single most common way a genuinely good process video dies.

3. The enumerated list

An explicit count in the opening, then rapid entries with visible checkpoints. Five tools. Three mistakes. Seven things nobody tells you.

The list is the most reliably average of the five, which is worth being blunt about. When we look at what separates the top of the list format from the bottom, the structural features we can measure barely distinguish them. The format itself is doing the carrying: the count is a contract, the contract creates a completion instinct, and almost any competent execution gets a floor.

That is a real advantage and a real ceiling. Lists are the safest shape to start with and the hardest one to break out with, because the thing that makes a list travel is item quality, which is not a structural lever anyone can hand you.

4. The reveal

A curiosity gap opened immediately and closed with something the viewer did not already know. Explainers, “here is why X actually happens”, myth corrections, direct-to-camera insight.

This is the shape most sensitive to hook craft, and it is where the largest execution gaps show up. Winning reveals open on a genuine unknown and land the payoff fast. Losing reveals open on a topic announcement, which is not a question, so there is nothing to stay for. If your videos live here, the five failure patterns in the hook mistakes that kill a video before second three are the highest-leverage thing you can read.

The other consistent separator in the reveal shape is shot length. Winners cut noticeably shorter and hit the point of maximum tension early. Reveals that meander are reveals nobody sees the end of.

5. The remix

Built on borrowed context: a reaction, a commentary layer, a stitched correction, a format everyone already recognizes with your material inside it.

The remix is the cheapest shape to produce and the most conditional. It works when the borrowed context is genuinely live and your addition is the point. It fails when the original is doing all the work, because then the viewer has no reason to associate the value with you. There is more on choosing what to borrow in duet, stitch, or just take the idea.

Every one of these shapes also fails, constantly

This is the part the “content types that go viral” listicles leave out, and it is the most important thing on this page.

When we look inside a single format class, the top of that class and the bottom of that class are the same shape. Both groups made a transformation video. Both made a list. The top group reaches many multiples of what that creator normally reaches, and the bottom group reaches a small fraction of it. The spread inside one shape is far larger than the spread between shapes.

So picking the right shape is not a strategy. It is a starting position. What separates the two ends of a class, over and over, is unglamorous: pacing, emotional charge, whether the promise is clear in the opening, and whether the video actually pays that promise off. Those are execution variables, and they are the ones you control.

The practical read: do not spend another week choosing a format. Choose one, then spend that week getting faster, clearer, and more committed inside it.

The shape that travels worst is the one everyone starts with

Personal narrative is the sixth shape, and it is the one we would tell you to be careful with early.

It works, but the winning versions of it lean on emotional intensity a stranger has no reason to grant you yet. A story shape on an account with an existing audience converts trust into retention. The same story shape on a cold account is asking a viewer to care about a person they met one second ago. It is not that personal content cannot travel. It is that it travels later, after one of the other five shapes has done the introducing.

If you are building from zero, take the shape that carries a payoff a stranger can evaluate in a still frame, and save the story for when the profile visit is already happening.

Mapping a shape onto what you actually know

The workable move is not “pick a topic that goes viral”. It is to take one shape and force your existing expertise through it. Three questions:

What in my work changes state? That is your transformation.
What do I make, fix, or assemble that has a satisfying finished object? That is your build.
What do people in my niche believe that is wrong? That is your reveal.

Most people can answer at least two of those in sixty seconds, and the answers are already inside their actual competence. That is the whole point: the shape is imported, the substance is native. This is also why we argue that a niche is one audience problem seen from many angles rather than one narrow topic, which we unpack in niche down or post everything.

If nothing comes, that is usually a research gap rather than an idea gap. Studying the tier directly above you, as we describe in study the tier above you, tends to surface the shape-to-substance mapping faster than staring at a blank notes app.

The two-week test that answers this properly

Pick one shape. Post it six times over two weeks with different subject matter each time. Hold the structure constant on purpose.

Then read the results against your own baseline, not against anybody else’s view count, because how many views counts as viral is a ratio question. If the six run wildly differently, the shape is fine and the execution variable is your lever. If all six land flat, you have learned something more valuable: that shape does not fit your material, and you have spent two weeks instead of two months finding out.

Repeat with a second shape. Two rounds of this is a real answer about your account, which is worth more than any general answer about the platform.

What this means for the next video

The honest version of “what type of content goes viral” is that no type does. Structures create the conditions for reach, and execution inside a structure decides whether reach happens. Anybody selling you a topic list is selling you the least transferable part of somebody else’s success.

Blossom scores the hook, names the format, and maps where a video builds and where it loses people, so the shape you are looking at stops being a guess. If you want to see which of the five your best videos already are, and where inside them the attention actually goes, start a workspace and run your last five uploads through it. And when you finish a video, give the ending the same attention as the opening, because the closing line is where the save is earned.

What Type of Content Goes Viral: Five Shapes, Not Five Topics