One video does three times what the others did. The obvious next move is to order five more like it, and the obvious next move produces five videos that all do slightly worse than average.

This happens so reliably that it is worth treating as a rule rather than as bad luck.

What you are copying is the surface

When a brand describes why something worked, it describes what it can see. The kitchen, the opening line, the format, the fact that it was filmed at night.

Those are the visible features. The cause is usually somewhere else, and it is often one of three things nobody wrote down: the video arrived at a moment when the audience was already interested, the creator did something slightly unusual that the feed rewarded once, or a small specific detail made it feel true and that detail was not in the brief.

Copy the visible features and you reproduce the shape without the reason.

What the brand thinks workedWhat often worked insteadWhat to ask for next
The formatThat the format was unfamiliar thenA different unfamiliar thing
The opening lineThat it named a real irritationAnother real irritation, said plainly
The settingThat it looked like the viewer's homeAny home that is genuinely someone's
The creator's energyThat they were not performingRoom for them not to perform again

The second row is the useful one to argue with internally. A line that worked because it named something true does not work again because it is the same line; it works again because the next line names something else that is also true.

What to actually reproduce

Copy the arrangement, not the flat

Reproduce the constraint, not the output.

If the winning video was made by someone filming in their own flat with no script, the reproducible thing is that arrangement rather than that flat. Give the next creator the same freedom, the same short brief, and the same permission to be unpolished, and you are running the same experiment rather than photocopying its result.

Ask the creator what they think happened

They watched the comments and you watched a dashboard, so they often know something you do not. This costs one message and it regularly changes the brief.

Keep one variable, change the rest

If you want to test whether the format mattered, keep the format and change everything else. Repeating all of it at once tells you nothing, because when the result drops you cannot say which part was carrying it.

When repeating genuinely works

Series formats are the exception, and they are worth knowing about.

If the first video was legible as an episode, something with a recognisable structure that can hold different content each time, then repetition is the point and audiences reward it. That is a different thing from copying a one off, and the tell is whether a viewer could describe the format to somebody else in a sentence.

Everything else has a short life by design. The most common mistake in this category is treating a good video as a template rather than as evidence about what your audience responds to.

Keep reading

Find UK creators →


Frequently asked questions

Should we book the same creator again?

Usually yes, and for a different video rather than the same one. The value in that relationship is that they now understand your product, which is worth more than the video you already have.

Can we run the winning video for longer instead?

Often the better move, and cheaper, as long as the licence allows it. Check the end date before you increase the spend rather than after.

How many attempts before we conclude something?

More than two, and with only one thing changing between them. Most brands stop at the point where the data is least useful.

What if the creator refuses to redo it?

Listen to why. A creator who says the second one will not land is usually right, and they are protecting your budget as much as their own reputation.