Never touch a winning video. Here's why.

The problem

A creator has a video that sells well, week after week. The brand wants more of that, so it asks for a “refresh.” The creator posts a new version: new opening, new cover, tighter first seconds, new music, new caption. Everyone treats it as the winner now.

Five changes, one new post, and two weeks later it sells half as well as the original. Nobody can say which change did it, and nobody went back to build on the original.

Every test is a comparison. A new hook is only “better” than something, and that something has to sit still while you look. The winner is already live and already earning. It does not need anything done to it. What it needs is to keep its job as the thing every new version is measured against.

“Touching” a winner means changing five things in the next version and calling that the new winner. Now you have five unknowns and one number. That is not a test. It is five guesses.

The example

The same five ideas, posted as five versions beside the winner, after two weeks.

Next round, V1 becomes the reference, and its next versions each change one thing. The music change goes in the notes as a thing never to do again on this product.

Do this in Verdict

1. 1
   Open Waves. A wave is one winner and the versions posted beside it. The card says Constant to protect at the top: the part every version keeps.
2. 2
   Each entry names what it changed, so two weeks later the chart above is already built.
3. 3
   When a creator or a brand wants to “refresh” a winner, log the idea as a new entry that changes one thing. Same idea, same afternoon, and you find out what it did.