Guide · Testing
How to Test 20 Video Ad Variations in a Day Without a Video Team
Big brands don't guess which ad will work. They make dozens of versions, run them all, and keep the ones that sell. Until recently that took a video team, a shoot and a week of editing.
Now a small brand can do the same thing in one day, with as low as one product photo or just with text.
This guide shows the exact workflow.
01Why 20 variations, not one “perfect” ad
Most ads fail, and nobody can reliably say in advance which one will win. Not you, not an agency, not a creative director. Social media platforms say the same thing in their own advertiser guides: give the algorithm several creatives and let it find the winner.
Small brands usually do the opposite. They spend two weeks on one polished video, it underperforms, and they conclude that “video ads don't work for us.”
The fix isn't a better single ad. It's more shots on goal.
02The rule: change one thing at a time
Twenty random videos teach you nothing. Twenty planned variations tell you exactly what your customers respond to.
Keep the product the same, and change one thing per variation:
| What to test | Example variations |
|---|---|
| Hook (first 2 seconds) | close-up of the product · the product in a hand · the problem it solves |
| Setting | clean studio · kitchen counter · outdoors in daylight · bathroom shelf |
| People vs no people | product only · a person using it · hands only |
| Mood | calm and premium · bright and energetic · cosy and warm |
| Format | 9:16 for Reels/TikTok · 1:1 for feed · 16:9 for YouTube |
Four hooks × three settings is already 12 ads. Add formats and you pass 20 quickly.
03The one-day workflow
Morning: plan (30 minutes)
- Pick one product and your best photo of it. A clear, well-lit photo matters more than the camera.
- Write down 4 hooks and 3 settings from the table above.
- Decide on your main format. Start with 9:16 if you advertise on Reels or TikTok.
Late morning: generate (2–3 hours)
- Upload the photo to Adreelo and describe the first variation in plain words, for example: “Close-up of the serum bottle on a marble bathroom shelf, soft morning light, slow camera push-in.”
- While it renders, write the next prompt. Keep everything the same except the one thing you're testing.
- Repeat until you have your set. Each video renders in minutes, so you can queue the next idea while you review the last.
Afternoon: reformat and review (1 hour)
- For your strongest videos, use Change aspect ratio in the player menu to get 1:1 and 16:9 versions without generating again.
- Cut anything that looks off: a warped product, odd hands, wrong colours. An ad that looks fake costs you trust.
Evening: launch
- Upload all variations into one ad set (Meta) or one ad group (TikTok) with the same budget and audience.
- Name each file by what changed (hook-hand_setting-kitchen_9x16) so the results read themselves.
04How to read the results
Give the ads 3–5 days and enough spend that each one gets real impressions before you judge it.
- Low watch time or CTR on every ad → the problem is the offer or the audience, not the creative.
- One hook clearly wins → make 5 new variations of that hook in different settings.
- One setting wins across hooks → that's your brand's visual world. Use it in your photos and website too.
Then repeat. Each round, you start from the last winner instead of from a guess.
05Common mistakes
- Changing everything at once. If the hook, setting and music all change, you can't tell what worked.
- Judging after one day. Ad platforms need time to learn.
- Only testing “pretty.” The raw, real-looking video often beats the glossy one, especially on TikTok.
- Stopping after one winner. Winning ads wear out. Keep a new batch going every couple of weeks.