The clearest way I have watched a brand waste money on creative is not by buying bad videos. It is a brand ordering three videos, running them as three ads, and calling that a test.
I understand why it happens. From where you sit, three videos is three chances for something to work. From where I sit, watching what comes back, three videos that differ in five ways each is three guesses with no way to learn anything from the result. If one wins, nobody can tell you why — so next quarter you are guessing again from the same standing start, and paying me again to produce the next three guesses.
That is a bad deal for you. It is also, frankly, a boring way for me to work. So here is how I think about volume, from the side of the table that has to actually shoot it.
You are split testing variables, not videos
The shift that makes everything else make sense: a creative test is not "which of these performs best." It is "which variable moves the needle for this product, with this buyer."
If three videos differ in hook, format, length, talent and call to action all at once, and one wins, you have learned that one specific combination of five things beat two other specific combinations. You cannot apply that to anything. It does not transfer to the next batch, the next product, or the next season.
If three videos share everything except the opening three seconds and one wins clearly, you have learned something durable about what stops your buyer scrolling. That insight outlives the ad. It goes into the next brief, and the one after that, and it compounds.
The difference between those two outcomes is not budget or talent. It is whether somebody structured the shoot around isolating a variable — which is a decision made in the brief, weeks before I pick up a camera.
The variables worth isolating
Roughly in order of how much they move performance:
The hook. The first two to three seconds. Consistently the highest-leverage variable in short form, and the cheapest thing I can give you variations of — often the same body footage with different openers cut onto the front.
The format. Talking head versus voiceover over B-roll, demo versus story, problem-first versus result-first. A structural difference, not a rewording.
The angle. Which benefit leads. Same product, same creator: convenience versus results versus price versus ingredients. This is where you find out what your buyer actually cares about, and more than once it has not been what the brand team assumed.
The talent. Real, but slower and more expensive to isolate, and easily confounded with everything else. Worth testing once the first three are settled.
Why ordering sequentially is the expensive way to buy from me
This is the part I most wish brands understood about my cost structure, because it would save them a great deal of money.
My costs are concentrated in the booking, not the file. Getting the space set up, the light right, myself ready, the product staged and myself into the headspace — that is the expensive part, and it is the same whether I then shoot one angle or six. Once I am set up, marginal variations are genuinely cheap.
So the brand that briefs six variants into one shoot day pays a fraction per asset of the brand that orders one video, waits three weeks to see how it performs, then orders another. That second brand is paying my full setup cost four separate times to end up with four assets — and taking four months to learn what the first brand learned in three weeks.
They are also, usually, the brand that messages me in a hurry, because sequential ordering means you are always discovering you need the next thing at the moment you need it live.
So how many?
What I would brief if I were sitting on your side:
Three to four hooks against one proven body. Highest return available, cheapest to produce. If you already have a video that works, new openers cut onto it cost very little relative to a fresh shoot.
Two to three genuinely distinct formats per product — structurally different, not reworded.
Two to three angles once you know the winning format, to find the message.
For one product going into a paid push that is roughly six to nine assets to read properly, which is exactly why single-video orders so rarely produce a clear answer. It is also why briefing several angles into one booking is dramatically more efficient than commissioning one at a time across a quarter.
Give each variant enough spend to mean something
Volume without budget behind it is just more guessing, and this is where I see good creative wasted.
If your test budget is spread so thin that each variant gets a few hundred impressions, the differences you observe are noise. Acting on noise is worse than not testing at all, because you will confidently brief the wrong thing next round and compound the error.
Give each variant enough spend and enough time to exit learning and produce a stable read. Fewer variants tested properly beats more variants tested badly, every time. If budget is the constraint, cut the number of variants — not the spend per variant. I would genuinely rather shoot you four assets you can read than nine you cannot.
Test before the season, not during it
Everything above has a deadline attached, and it is earlier than most plans assume.
Testing in October buys information at ordinary prices. Testing during a shopping season buys the same information at peak auction rates, in the weeks when your margin is thinnest, using budget that should be scaling a proven winner rather than finding one.
Which is the real reason my calendar matters to you. Brands book me for a date and think about the delivery deadline. The deadline that actually governs is the one where you have finished split testing and picked your winner — and that sits weeks ahead of when the campaign goes live. If you want to be scaling a proven asset on Black Friday, the booking conversation happens in September.
Creative fatigue is why this never ends
The other reason volume matters: winners decay. Frequency climbs, your audience has seen it, acquisition cost drifts up. Normal, expected, not a sign the content was bad.
Brands that treat UGC as a project — commission a batch, run it until it dies, panic — live permanently in that panic, and the panic is usually the point at which they message me. Brands that treat it as a pipeline always have the next thing warming up while the current winner still works. That second group is calmer, spends less per asset, and gets better creative, because each round is informed by what the last one taught them.
What this changes about how you commission
You stop ordering "a video" and start ordering a test structure — a set of hooks, a couple of formats, a defined set of angles. You tell me what you are trying to learn, not just what you want shot, and I will structure the day around producing a readable result instead of a nice one.
You also make sure the rights cover paid usage across the whole batch from the start, because the entire point is putting spend behind whatever wins, and you do not know yet which one that is.
And it is worth being honest about which tool you are reaching for. UGC and influencer marketing solve different problems — a testing programme is squarely a UGC exercise, because you need volume, variation and full control of distribution, none of which you get from renting somebody's audience.
The brands whose fifth round of creative is better than their first are not the ones spending the most. They are the ones who built something that learns — and who worked with someone who understood that their shoot day and your media plan are two halves of the same problem.