Blog · Guide · · Updated
Why we sell AI ads in packs of five (and how to test them on Meta)
Short answer
Because one ad is a guess and five ads are a test. Five ads with the same offer and five different opening angles show which idea your buyers respond to. One ad only tells you whether that one idea worked.
Most small brands we meet have run exactly one video ad. It didn’t do much, and the conclusion was “Instagram ads don’t work for us”. That’s rarely the right conclusion. Usually the product and the offer were fine. The first three seconds just didn’t land, and nobody got to see what a different opening would have done.
Why does one ad rarely tell you anything?
If one ad flops, you don’t know whether the problem was the hook, the feeling, the product shot, the offer or the audience. You’ve spent the budget and learned very little.
Five ads that keep everything the same except the opening angle give you a comparison. When one holds attention and another doesn’t, the difference is the angle, and that lesson carries into every ad you make afterwards.
What should change between the five ads?
Only the angle. The product, the offer and the call to action stay exactly the same across all five, so the test is fair. Here’s a pack we’d write for a Chennai brand selling a ragi snack:
| Ad | Angle | Opening hook |
|---|---|---|
| 1 | Craving | A slow close-up of the crunch, sound up |
| 2 | Relief | The 4 pm office slump, and a snack you won’t regret |
| 3 | Pride | A grandmother’s recipe, now in a pack |
| 4 | Proof | The ingredient list, read out in eight seconds |
| 5 | Belonging | One pack opened on a desk, six hands reaching in |
Notice what doesn’t change: the pack shot at the end, the price, the “order now” line. Change those too and you can’t tell what made the difference. This is the core of Honey Badger Marketing: find the feeling that sells, then go after it from five sides.
Which formats does Meta need?
Meta video ads need a vertical 9:16 cut (1080 × 1920 pixels) for Reels and Stories and a 4:5 cut (1080 × 1350 pixels) for Feed, where it fills more of a phone screen than a square does. Every Spilanth ad ships with both cuts. A few details matter as much as the ratio:
- Keep text away from the edges. In Reels and Stories the top and bottom carry the account name, captions and buttons.
- Write for the sound off. Captions carry the message and the sound is a bonus.
- Land the hook early. The first three seconds decide whether anyone sees the rest.
How do you read the results?
Run all five at the same time, with the same budget, audience and objective. Then look at three numbers for each ad in Ads Manager:
- How many people watch past the opening. Many advertisers divide three-second video plays by impressions and call it the hook rate.
- Click-through rate. Whether the ad made someone want to know more.
- Cost per result. Purchases, leads or messages, whatever you’re optimising for.
Give the test time. Judging after one day or a few hundred impressions is a coin toss. When the numbers disagree, that’s useful too: if one ad has the best hook rate and another the cheapest result, the next ad can combine them.
What happens after the test?
Keep the winner running and build the next pack around it. If “Craving” held attention and “Proof” got the clicks, the next five might all open on a craving hook and end on proof. That’s how an ad account gets steadily better instead of starting from zero every month.
Frequently asked questions
How many ad creatives should a small brand test on Meta?
Five is a practical number for a small brand: enough variety to see which angle works, and few enough that each ad gets a fair share of the budget. Keep the offer the same and change only the opening hook.
What aspect ratio should Meta video ads be?
Use 9:16 (1080 × 1920 pixels) for Reels and Stories and 4:5 (1080 × 1350 pixels) for Feed. Every Spilanth ad ships with both cuts.
What should change between test ads?
Only the angle, meaning the opening hook and the feeling it appeals to. Keep the product, offer and call to action identical so any difference in results comes from the idea.