Test your hook, format and call to action in one round. See which element moves results, and which elements work best together.
One ad for every combination
The basics
A factorial test is an experiment that tests several ad elements, or factors, at the same time. It shows the effect of each element on its own, and how the elements work together.
Say you want the best mix of hook, format and call to action for a video ad. Each element gets two options. That is a 2 × 2 × 2 factorial test.
You make one ad for each combination and run all 8 at the same time.
| Factor | Option 1 | Option 2 |
|---|---|---|
| Hook | Question | Bold statement |
| Format | UGC | Product demo |
| CTA | Shop now | Learn more |
The answers
One round of 8 ads answers three questions.
Does a question hook beat a bold statement on average? Compare the 4 ads with a question hook against the 4 with a bold statement.
Does the question hook work well with UGC but poorly with a product demo? Compare the hook result inside the UGC ads with the result inside the demo ads.
Which of the 8 ads gives the best result, for example the lowest cost per purchase? That is the ad to scale.
Worked example
Example numbers for cost per purchase, not from a real account. Lower is better.
Average the 4 ads that share each option.
The question hook saves $6 per purchase on average. The hook matters most here.
Compare the two hooks inside each format.
Inside UGC
Inside product demo
The question hook saves $12 in UGC and nothing in demos. It only works with UGC.
Ad 1 wins: question hook, UGC and Shop now, at $22 per purchase.
The next round keeps the winning pair, question hook with UGC, and tests new elements on top of it.
Why it is useful
A normal A/B test changes one element at a time. A factorial test also shows how creative elements work together. A full factorial still tests every combination. The time you save is against running one A/B test after another.
Pick your factors
Connecty breaks down every video ad second by second and labels these elements. Each one can be a factor in your test. After the test, the same labels show what the winning ad did differently.
Question, pain point, benefit, curiosity, social proof, offer first, demo first
UGC, talking head, product demo, testimonial, motion graphic
A face, the product, text on screen, or a demo
Testimonial, demo, measurable claim, before and after, reviews on screen
Early, middle or late in the ad
The second the ad makes its main point, and how many viewers still watch then
No offer, a visible offer, or a clear and prominent one
Direct buy, soft, learn more, urgency
Curiosity, pain, excitement, aspiration, humor
Lo-fi UGC or polished
Keep each factor to two options. 3 factors make 8 ads. 4 factors make 16.
Test ideas
Each category comparison ends with concepts worth testing. Here is how three of them turn into one 2 × 2 × 2 test.
7 of 21 ads say vets back the food, and no vet speaks in any of them. Only 2 of 21 ads show a price.
11 of 34 videos promise a cooler night, and none gives a number. 28 of 34 make no offer.
17 videos say clinically proven, studied or shown, and few give the size of the result. Only 18 of 104 have a clear call to action.
More categories in the video ad research.
Step by step
Smaller budget
8 ads split your spend 8 ways. With a small budget, run 4 of the 8 instead.
Pick Ad 2, Ad 3, Ad 5, Ad 8. Every option shows up in 2 of the 4 ads, so you still see the effect of each element on its own.
The 4-ad version cannot separate the pairs. Confirm the winning pair in your next round.
Questions
A factorial test is an experiment that tests several ad elements at the same time, such as the hook, the format and the call to action. You make one ad for every combination. The results show the effect of each element on its own and how the elements work together.
Multiply the options. Three elements with two options each need 2 x 2 x 2 = 8 ads. Four elements need 16. With a small budget, a half factorial with 4 of the 8 ads still shows each element's own effect.
An interaction effect means one element works differently depending on another. For example, a question hook may beat a bold statement in UGC ads but make no difference in product demo ads.
A normal A/B test changes one element at a time, so three elements take three rounds and never show how they combine. A factorial test answers all three in one round and shows which pairs work together. A full factorial still tests every combination.
Connecty analyzes every video ad in your account, second by second: hook, proof, offer and the key deciding moment. Try it free for 14 days.
Continue with GoogleSign up with emailYou connect your Meta ads account after you sign up.Made by Connecty, which reads and compares video ads. See every category in the video ad research.
