To test ad creatives on a small budget, change one thing at a time (the hook, the angle or the format), run three to five versions side by side with the same audience and budget, give each one enough spend to get a fair result, and judge them by cost per purchase, not likes or clicks. Keep the winners running, switch off the losers, and add a few new creatives every week, because even great ads stop working after a while.
Why creatives matter more than targeting now
Meta and TikTok's systems are good at finding buyers if you give them a broad audience and a purchase signal from your pixel. What they can't do is make a boring ad interesting. Today the creative (the video or image and its words) does most of the targeting: different creatives reach different people.
Before you test
- Pixel on your store, reporting purchases. On Paylead, paste your Meta and TikTok pixel IDs in Store Settings → Integrations, and add a Conversions API token. Connect your pixels.
- A page that converts. Testing ads to a weak page wastes money. See why your store isn't making sales.
- A budget you can afford to learn with. For example, GH₵ 50–GH₵ 100 per ad per day for 4–7 days.
What to test, in order
| Order | Test | Example |
|---|---|---|
| 1 | Angle (why buy) | Save money vs save time vs status |
| 2 | Hook (first 1–3 seconds) | A question vs a bold claim vs a result |
| 3 | Format | Talking to camera vs product demo vs customer review vs carousel |
| 4 | Offer | Price alone vs with bonus vs bundle |
| 5 | Details | Captions, thumbnail, first line of text, length |
Angles and hooks make the biggest difference. Button colours and tiny wording changes rarely matter.
How to set up a test
- One campaign, goal: Sales, event: Purchase.
- One broad audience (country or cities, age range).
- 3–5 ads, each different in only the thing you're testing.
- Equal chance: either one ad set per creative with the same budget, or let Meta spread spend across ads and accept it'll favour some.
- Don't touch it for at least 3–4 days.
Reading results
Look at cost per purchase first, then these to understand why:
| Metric | Tells you | If bad |
|---|---|---|
| Hook rate (3-second views ÷ impressions) | Does the start stop the scroll? | New first line or opening shot |
| Click-through rate | Is the ad interesting enough to click? | Stronger angle or offer |
| Landing page views → purchases | Does the page deliver what the ad promised? | Match the page to the ad |
| Cost per purchase | Is it profitable? | Compare to your profit per sale |
Example (GH₵ 200 product, ~GH₵ 190 after fees):
| Ad | Spend | Purchases | Cost per purchase | Decision |
|---|---|---|---|---|
| A: "Stop wasting money" hook | GH₵ 400 | 9 | GH₵ 44 | Keep, raise budget slowly |
| B: Product demo | GH₵ 400 | 3 | GH₵ 133 | Off |
| C: Customer review | GH₵ 400 | 6 | GH₵ 67 | Keep, try new hook |
When is a result real?
With small budgets, a couple of lucky sales can mislead you. Rough rules:
- Wait for at least a few purchases per ad before deciding.
- If two ads are close, keep both and test something bigger next.
- If an ad spends 2–3 times your target cost per purchase with no sale, turn it off.
Keep a creative pipeline
Ads wear out as the same people see them. Plan to add 3–5 new creatives a week or two:
- New hooks on your winning video.
- New angles from marketing angles.
- Customer videos and reviews.
- Behind the scenes, unboxing, before/after.
Phone-filmed, natural videos often beat polished ones on TikTok and Reels.
Frequently asked questions
Should I test on TikTok or Facebook first?
Wherever your buyers spend time. Younger audiences lean TikTok; many working adults and business owners are reachable on Facebook and Instagram.
How much should I spend testing?
Only what you can afford to learn with. A week at GH₵ 150–GH₵ 300 a day across 3 ads is a reasonable small test.
Read next: 50 ad hooks that stop the scroll and how to track sales from ads.
