The VSC framework

The filter in one picture

The order matters. Viral is cheap to check, so it goes first. Convertible needs you to read comments, so it goes last, once an idea has earned the time.

Most swipe files only check V. They are folders of videos that got big numbers. VSC is V and S and C, which is a much smaller and much better set of ideas.

V - Viral: is it a real outlier?

Views on their own lie. 800K views looks huge, until you see the account usually gets 4M. A 40K video on an account that usually gets 2K is the stronger signal, because something in that video made it travel.

So Viral is two tests, and an idea has to pass both.

• It's an outlier. Divide the video's views by the account's median recent views. Around 10x is a strong outlier. 100x and up means something structurally different happened. Under 2x is noise: a big account posting a normal video.
• It repeats. The same format has worked for more than one creator, ideally small accounts. My first version of the rule was 100K+ views on several similar videos from accounts under 5,000 followers. If it appears once, ignore it.

What the data says

Handler keeps a library of TikToks that founders and our team flagged as winners: 5,896 videos from 2,402 accounts, each scored against its own account's baseline. The median one has 254K views and sits at 75x its account's normal.

Here's the part that matters for Viral. 1,052 of the 4,671 real outliers in that library have fewer than 100,000 views. Sort by views and you throw away more than one in five of the best ideas, because they came from small accounts.

Fake signals

• One viral post, no copycats from smaller accounts.
• A huge account having a normal day.
• Lots of likes, few saves or shares.
• A trend tied to one sound that dies when the sound does.

S - Scalable: can someone else make 30 of these?

A viral video tells you something worked once. A lucky night doesn't grow an app. A format you can run every week does.

We ask three questions. A no to any of them means the idea isn't worth farming.

1. Can it be made 30 times without the audience getting bored?
2. Can a creator who isn't you make it from a brief, without your taste?
3. Does it work without one particular personality?

A scalable format is a locked structure with swappable parts. We lock the slides, pacing and emotion, and only change the hook text, the first frame and the phrasing. Then we batch 10 to 30 versions and spread them across creators and accounts.

Handler's library backs this up. Of the 4,671 real outliers, 3,905 (84%) are formats where nobody has to perform to camera: silent UGC, app demos, slideshows and text-on-screen videos. Talking heads can win, but they depend on one person, which makes them the hardest to scale. Phone call videos are a good scalable example: one setup, a new script every day.

C - Convertible: do the comments want the product?

10M views with no reason to download is worth nothing. Views aren't installs. A viral format can pull in exactly the wrong audience: people who came for the joke and will never open your app.

The comments tell you which one you've got.

• Converts. "what app is this", "where do I get this", "how do I do this", "this explains my situation", people tagging a partner or friend.
• Doesn't. Jokes, memes, "so true", fire emoji, people reacting to the sound.

What the data says

Handler reads the comments on every video in its library and flags whether they show interest in the product. Of the 4,671 real outliers, 1,978 (42%) failed. They were genuinely viral, and the comments still had nothing to do with an app.

Bigger doesn't fix it. Among outliers with a million views or more, 643 of 1,304 (49%) failed Convertible. And the format changes the odds a lot.

App demos convert best because the product is the story. Silent UGC wins plenty of views, but half the time the audience is there for the feeling, not the app. That doesn't make silent UGC bad. It means the hook has to point at your product, or the views go nowhere.

For apps the goal isn't the click. It's someone who installs knowing what they're getting, so they activate and stay. That's why every brief is written around the moment the app clicks for a new user, not the trending sound of the week.

How VSC runs on our board

At Agniverse Media, VSC sits at the top of a fixed pipeline. Angles are who we're talking to and why they'd care. Hooks are the first three seconds, written as bets against each angle. Scripts are what the creator shoots. Creators work from a live brief that updates before every shoot.

Every angle gets 3 to 5 hooks, and every hook gets 5 to 10 videos, because TikTok first shows a post to a few hundred people and one post on its own can just be unlucky. Then the numbers decide.

• Scale. Two or more videos past 10,000 views. Make more, same structure.
• Mutate. About one in three videos past 2,000. Change one thing and run it again.
• Kill. Neither. Drop it and move the budget to the next bet.

Videos under 200 views don't count toward the read. That's usually the account, not the hook.

Across our own apps and client apps, the board holds 84 angles, 522 hooks and 7,377 videos. Of the 87 hooks that have reached a decision so far, 38 were killed, 28 mutated and 21 scaled. On Flamme alone it was 9 angles, 34 hooks and 4,228 videos.

VSC doesn't make every idea win. Roughly one decided hook in four gets scaled. What it does is make the losers cheap, because nothing reaches a camera without passing three checks first.

Watch the walkthrough

Chapter 13 of my TikTok farming playbook covers VSC end to end: real versus fake viral signals, the three scalability questions, and how to read comments for intent.

Run it on your own niche

1. Find 20 videos in your niche that look like they worked.
2. Score each one against its own account's median views. Drop anything under 10x.
3. Keep only formats you can find on at least two creators.
4. Ask the three Scalable questions. Drop anything that needs one person's personality.
5. Read the top comments. Keep only the formats where people ask about the product.
6. Film 5 to 10 versions of each hook, then kill, mutate or scale on the numbers.

Step two is the hard one by hand, because you need each account's baseline. Handler's free TikTok Outlier Calculator does the math, and its VSC guide goes deeper on validating a single idea before you build anything.

If you'd rather have it run for you, that's Agniverse Media. For the wider playbook around it, read Organic UGC 101.