The problem
At small volume a strategist can watch every video, write a diagnosis, and decide what to make next by hand. At fifteen hundred videos a month, that breaks. There is not enough time in a month to watch fifteen hundred videos, let alone think carefully about each one.
So the job changes shape. Instead of analysing every video, the strategist runs a system that finds the handful of videos worth their attention and turns those into the next round of scripts. And because a brand has a revenue number it needs to hit on a fixed budget, every test has to do two jobs at once: make money now, and teach you something that makes more money later.
Every video still gets its numbers pulled automatically: creator, product, hook, angle, format, spend, revenue, the funnel, and how many people actually saw it. But only a small slice deserves a strategist’s time. Of fifteen hundred videos in a month, most never get enough views to mean anything. A few hundred give a preliminary read. A smaller group shows something worth a real look. Only a few dozen produce a decision that changes what gets made next. The exact counts don’t matter. The shape does.
Making money and teaching you something are two different questions
Ranking everything by revenue alone is a mistake. Three videos, same month.
| The video | The revenue read | The learning read |
|---|---|---|
| Clean winner, big return on real spend | Obvious. Make close versions now. | Teaches little on its own. You don't yet know which piece carried it. |
| Modest revenue, strong Payoff and Click, weak Convert | Looks mediocre. | Worth more than the winner. It hands you a narrow problem: the video works up to the click, and something after it is missing. |
| Almost nothing, on almost no spend | Looks like a loser. | Teaches nothing yet. It never got enough distribution to prove anything either way. |
So the real question is not just whether it made money. It is whether it taught you something you can act on. And calling that third video a loser is the most expensive mistake on the page, because you will file a script as dead that was never tested.
The example
At this volume a dashboard full of numbers is overwhelming. What you actually need is a short list of decisions to make today.
| Verdict | What it means | What you do |
|---|---|---|
| SCALE | Strong all the way through | Make several close versions now, while it is working. |
| OPTIMIZE | Strong through Click, weak on Convert | Keep the hook and body. Test the proof and the offer. |
| REWORK | Weak at Hook and Payoff, converts well among the few who stay | The offer works. The opening doesn't. Replace the first few seconds. |
| HOLD | Not enough views yet | No verdict. Don't call it a failure. |
| KILL | Weak at every step, with enough views to be sure | Stop putting production slots behind this angle. |
That turns the job from reading content into working a queue.
The target decides how much room you have to experiment
A brand with a revenue number and a budget has a return it needs overall. That number should decide how much of your production goes to new ideas versus close versions of what already works.
| Where the account sits | How to spend the slots |
|---|---|
| Running above the target | Room to try new hooks, new creators, new angles. The current winners can absorb some experiments that don't pay off. |
| Right on target | Keep a mix. Most production stays close to what is proven, a smaller share tests something new. |
| Running below the target | Get conservative. Not “what is a new idea” but “what is the shortest path from something already working to another video that will sell.” |
Tests are not all equal in how far they stray from proven ground, either.
| Size of the change | What moves |
|---|---|
| Smallest | Same creator, same script, same visuals. New hook only. |
| Small | Same hook and angle, different proof. |
| Bigger | Same underlying idea, new creator and new script built around it. |
| Biggest | New audience, new message, new format entirely. |
When revenue is under pressure, stay close. When there is room to spare, the bigger swings are worth taking.
Group videos by what they are testing
At fifteen hundred videos, thinking “I have fifteen hundred videos” is the wrong frame. The better frame is “I have a few dozen ideas being tested across fifteen hundred executions.” Once videos are grouped by the idea they test, a hook style, a proof format, an angle, you can compare across the group instead of one video at a time. That is a much stronger finding than any single video’s number, and it changes hundreds of future decisions at once rather than just the next one.
It also matters to separate what got attention from what got the sale. A hook style might reliably win Payoff without moving Convert at all. Confusing the two means crediting a hook for a sale that came from the offer, or scrapping a hook that was doing its job before something later in the video lost the sale.
Write it down, or you start over every month
Every real test should leave behind a rule, how much you have seen, what happened, and what to try next.
| What we believe | How much we have seen | What happened | Next test |
|---|---|---|---|
| Testimonial hook beats claim-only hook | Most of a batch | Payoff up meaningfully | Try three new testimonial styles |
| Price anchor raises Click | About half a batch | Click up | Test placing it earlier against later |
| Long ingredient explanation helps | A handful | No change | Stop spending slots on it |
Without this, every new month starts from memory, and nobody remembers the outcome of three hundred old tests. With it, month two starts with everything month one already proved.
Every video costs creator time, editing time and a strategist’s attention, on top of any media spend. So the question is not how many videos you can make. It is which briefs produce the most revenue and the most learning per slot. A batch of brand-new ideas might return a fraction of what a batch of close versions of a proven winner returns. Once that is visible, the production plan stops being “come up with more ideas” and starts being “spend slots where they pay off.”
So the job becomes this. Notice where the account is leaking or winning. Work out what inside the creative is causing it. Write a testable guess. Turn that into a small batch of controlled videos. Check the result against what you expected. Then go again with what you just learned added in. The advantage is not being able to personally judge fifteen hundred videos. It is running a system that finds the few worth learning from, and making sure the lesson survives into next month.
Do this in Verdict
- 1The engine already sorts every video into SCALE, OPTIMIZE, REWORK, TEST, HOLD or KILL. That is the decision queue in this lesson, built into the app rather than something you assemble by hand. Tomorrow's list, on Plan, is that queue for today.
- 2The Leak Finder flags videos worth rescuing even when their revenue looks unremarkable, and marks the reusable ones, so a video that is strong upstream and weak at Convert doesn’t get thrown away.
- 3
- 4The Pre-Slot Screener and its Window Score read the budget and the room, which is what decides how aggressive a test should be.
- 5Every guess and result goes in Learnings, so the account’s knowledge carries into next month instead of resetting.