Knowing whether your videos work. Here are the indicators that matter by objective, how to attribute results and a simple test structure.
Which indicators really matter
Seven measures, depending on the objective.
Three-second retention, to judge the hook.
Fifteen-second retention, to judge the body.
Cost per click, to compare creatives against each other.
Acquisition cost, the only indicator of real sales.
The add-to-basket rate, for a product page.
The creative's lifespan, before it wears out.
And the cost per video actually run, on the production side.
What not to measure
Six figures that mislead more than they inform.
Views, which mostly depend on budget.
Likes, unconnected to purchase.
The number of comments, without reading them.
Reach, which says nothing about conversion.
Overall engagement rate, too aggregated.
And a whole campaign's performance, which hides the gaps between creatives.
How to attribute results
Six rules that make attribution possible.
Run one video against a non-native creative, in the same group.
Keep the same budget for each variant.
Change only one thing at a time.
Name every file to the convention, dealt with above.
Track by variant, never by campaign.
And let it run long enough to clear the noise.
What a file name must contain
Seven elements, separated by a simple dash.
The creator's name, in three letters.
The product category.
The concept, in one word.
The hook type.
The format, vertical or horizontal.
The delivery date.
And the version number.
What test structure gives reliable answers
Six decisions for an honest test.
Three creatives minimum, never one.
An identical budget per creative.
A single variable that changes.
Two full weeks of distribution.
No modification along the way.
And a written decision before launching, on what will trigger a stop.
How to calculate the cost side
Seven items, to reach a real cost.
The price paid to the creator.
The usage rights, invoiced separately.
The cost of videos never run.
The internal time for briefing and review, in hours.
The cost of replacing a creator who failed to deliver.
The platform or agency fees.
And the cost of delay, when there is one.
When creative fatigue distorts your figures
Six signs that a creative is wearing out rather than failing.
The acquisition cost rises week after week.
Three-second retention drops first.
Exposure frequency rises on the same audience.
The click rate degrades before the sales.
The comments become repetitive.
And a new creative immediately starts better.
What to do when a creative wears out
Six actions, from the cheapest to the heaviest.
Replace the first three seconds.
Change the first line of subtitles.
Re-edit from the existing rushes.
Test the same video with no music.
Order a variant from the same creator.
And only then, order a new shoot.
The simple signs that a video works
Six signals readable with no complex tool.
It holds its acquisition cost over several weeks.
It withstands a budget increase without degrading.
It works on two different audiences.
It produces comments asking product questions.
It stays usable on a product page.
And it survives a change of season.
Can you measure without paid advertising
Yes, but with other markers. Organically, you look at retention, the number of messages received mentioning the product, the trend in brand searches, and the add-to-basket rate from the page where the video is embedded. It is less precise, slower, and it needs a longer comparison period.
How long to let a creative run
Six duration markers, before concluding.
Never less than seven days, whatever the budget.
Two weeks for a decision to stop.
Three weeks if the product has a long buying cycle.
One month to judge a product page.
Longer if the daily budget is small.
And always the same duration for every variant compared.
What the table must contain
Seven columns, not one more.
The file name, to the convention.
The creator.
The concept and the hook type.
The launch date.
The budget spent.
The result, as acquisition cost.
And the status: live, winner, stopped.
How to compare two creators
Six conditions for the comparison to mean anything.
The same brief, word for word.
The same product.
The same distribution budget.
The same audience.
The same period.
And a single variable that changes, the person.
What distorts a comparison
Six method errors, frequent and invisible.
Different budgets between variants.
Staggered launch dates.
An audience modified mid-test.
A creative running on two platforms at once.
A season changing in the middle of the test.
And a file renamed along the way.
How to start measuring this week
Seven steps, in order.
Write the naming convention, in one line.
Rename the existing files.
Create a table with one row per creative.
Add columns for cost, result, launch date.
Set an identical budget per creative.
Launch three creatives in parallel.
And touch nothing for two weeks.
Who should keep the table
Five rules so measurement survives team changes.
One responsible person, named.
A shared file, never a local spreadsheet.
A weekly update, on the same day.
A row written even when a creative is stopped.
And a monthly review with someone else, not alone.
What is measured, and over what period
| The question | The metric | The window |
|---|---|---|
| Does the video hold | 3-second retention | 48 h |
| Does it sell | Cost per purchase | 7 to 14 days |
| Does it last | The day the cost climbs back | 3 to 6 weeks |
| Does it reduce returns | The product return rate | A full cycle |
What skews the calculation
Comparing two different periods, changing the budget mid-test, or crediting a sale to the last video seen. Each of these produces an indefensible figure.
What almost nobody counts
The cost avoided. A video explaining post-purchase use lowers returns and support requests, and that saving appears in no campaign table.
The mistakes that distort everything
Six reflexes that make measurement useless.
Changing two things at once.
Stopping a test before two weeks.
Comparing creatives on different budgets.
Judging on views rather than acquisition cost.
Failing to distinguish wear-out from failure.
And losing track of the original file.
One last marker: measurement does not need a sophisticated tool, it needs discipline on three points, a file name, an equal budget, and a single variable that changes.
Sources
- Portal do Empreendedor (MEI), Brazilian government
- Receita Federal, Brazilian tax authority
- Simples Nacional, official portal
Checked on 4 September 2026. Thresholds and rates change: where this guide and the official source disagree, the official source is right.
