The takeaway: Rather than betting on one video, split your ad into three parts: the opening (Hook), the middle (Body), and the closing (CTA). Write 5 openings, 3 messages, and 2 closing lines to get 5 × 3 × 2 = 30 different videos from the same idea.

What are the three parts of an ad in a testing matrix?

  • Hook (the first 3 seconds): Determines whether viewers stop or keep scrolling.
  • Body (the middle): Explains the problem, the solution, and why this product.
  • CTA (the closing): Tells viewers what to do now: order, try it, or use the code.

What does the matrix look like for a perfume store?

5 openings:

A woman browsing perfumes on her phone surrounded by perfume bottles shopping bags and content lists showing the ad matrix
A shopper browsing perfumes on her phone within a perfume store content matrix · AI illustration
  1. "This perfume had everyone in the majlis asking what I was wearing."
  2. "Don't buy oud before you see this."
  3. "Costs less than designer brands and lasts longer."
  4. "The Eid gift you can't go wrong with."
  5. "I tried 10 perfumes this month… this is the winner."

3 messages: Longevity, affordable luxury, and suitability as a gift.

2 closing lines: "Order now with free delivery," "Use the code for a discount."

How do you read the results?

  1. Start with the openings: compare the percentage who watched the first 3 seconds (Hook Rate) and keep the top two.
  2. Then the messages: look at completion rate and clicks.
  3. Then the closing: compare cost per order (CPA).
  4. Combine the winners into new versions and keep going.

Why does this method matter now?

Meta Advantage+ and TikTok Smart+ algorithms find the right audience for each ad themselves. The more versions and variety you give them, the faster they find people who buy.

How do you start your first testing matrix?

Start small: one sales angle, three openings, one body and one call to action. That is only three versions, and it shows which opening stops the scroll. Then lock the winning opening and change the body, then the call to action. Changing one stage at a time tells you why results improved.

What are the common ad testing mistakes?

  • Changing the opening, body and call to action at once, so you cannot tell what worked.
  • Stopping an ad before it gathers enough data.
  • Testing openings so similar that the idea does not change.
  • Comparing versions across different audiences or budgets.
A confused man scratching his head while looking at tangled phones shopping bags charts and scattered papers
Confusion over scattered phones and papers reveals common ad testing mistakes. · AI illustration

How do you set a budget for the testing matrix?

Split the budget equally across versions at first so each gets a fair chance. With a small budget, test fewer versions per round instead of spreading a small amount over 30 versions, because a version that does not gather enough data gives you no clear verdict.

Small
What to test in round one
Hooks only: 5 hooks for one message
Medium
What to test in round one
Hooks with two messages
Large
What to test in round one
The full matrix, in stages

How do you carry the winner into the next round?

After each round, keep the winning part and change only one part. If one hook wins, lock it in and test new messages with it. This way you build the ad piece by piece and know the reason behind every improvement instead of guessing.

  1. Lock in the winning hook.
  2. Test new messages with it.
  3. Lock in the winning message, then test calls to action.
  4. Start a new cycle with new hooks when the ad tires.

How do you name matrix versions so the results are easy to read?

Give each version a short name that shows its parts, such as R1-H3-B2-C1: round one, the third hook, the second message and the first closing line. When your ad testing matrix reaches 30 versions, that name becomes your guide in the report, so you can see which hook or message keeps appearing in the winners without opening every video.

R (round)
Values
R1, R2, R3
What it tells you
Which test round the version launched in
H (hook)
Values
H1 to H5
What it tells you
Which hook stopped the scroll
B (message)
Values
B1 to B3
What it tells you
Which message kept viewers watching to the end
C (closing)
Values
C1 and C2
What it tells you
Which closing line brought orders at a lower cost

For what the metrics mentioned here mean and how they are calculated, see the digital ad terms glossary.

How do you test a new product's angles before paying for a shoot day?

Start with the same matrix, but with versions built from product photos instead of filming: choose three selling angles, write several openings for each, then test them on a small budget. The angle that proves itself then earns more versions, or a shoot day if it needs a real experience, so the big production budget goes to an idea you tested rather than a guess.

  1. Prepare clear product photos from several angles and the product page link.
  2. Write three selling angles, with several openings each in the dialect your audience speaks.
  3. Produce the versions in 9:16 and launch them on an equal test budget.
  4. Carry the winning angle into the next round, or into a shoot day if it needs a real experience.

With AI production the first batch arrives 48 to 72 hours after script approval, and to see the idea on your product before paying anything, request a free sample video.

Frequently asked questions

How long should a test run before judging a version? Until it gathers enough data for a clear decision; the time depends on budget and audience size.

Should I test on Meta and TikTok together? Yes, but read each platform's results separately, because audiences behave differently.

What do I do with losing versions? Stop them, and keep what they taught you for writing the next openings.

Does the matrix work for small accounts? Yes, with fewer versions per round, locking in the winner and changing one part at a time.

Who writes the hooks and messages? We write them in the right dialect for your audience, and you approve them before production.

Should I test dialect inside the same matrix? Test it in a separate round after locking in the winning version, for example comparing Najdi with general Gulf Arabic on the same script, so you see the effect of dialect alone.

What do you need from me to prepare the first matrix? Only product photos and a store link, with no sample shipping, and the first batch arrives within 48 to 72 hours of script approval. You can also request a free sample video before paying.

Part of our approach to Scaling Ad Creative from 30 to 200 Videos per Month. Try the Gulf hooks bank.