WhatRunsIndia"Test lots of creatives" is advice every brand hears. Then someone uploads fifty ads into one ad set with ₹2,000 a day of budget, and the system spends almost everything on three of them while the rest get a few hundred impressions each. Nothing is learned. Here is a structure that makes a large number of variations work on a modest budget.
Why not launch all 50 at once?
Meta's delivery system concentrates spend on the ads it predicts will perform, often very early. With a small budget and many ads, most variations never get enough delivery to prove anything. You end up with a few lucky winners and lots of untested ideas.
The fix is to test in rounds, moving from big differences to small ones.
Round 1: which concept wins?
Start with five clearly different concepts for one proven product. Different scenes, different people, different reasons to buy. Our guide on testing concepts instead of colours explains what counts as a concept.
- Budget: enough for each concept to get some purchases in about a week.
- Signal: cost per purchase, with click-through rate as an early hint.
- Outcome: one or two concepts clearly ahead.
Round 2: which execution of the winner wins?
Now make 10 to 20 variations of each winning concept. Keep the idea and change the execution:
- Model or person: different age, look or styling.
- Framing: full length, half body, close-up on the detail.
- Background: same type of setting, different location or time of day.
- Headline: three or four different lines on the same idea.
- Format: 4:5 feed, 1:1 square and 9:16 story versions.
Run these in groups of five or six rather than all at once, so each gets real delivery. Pause clear losers after a few days and add the next group.
Round 3: keep the winners fresh
By the end of round 2 you have a small set of proven creatives. Those will tire. Keep the remaining variations in reserve and swap them in as frequency rises and cost per purchase creeps up. Over a month or two, that is how you end up running fifty variations without ever wasting budget on fifty at once.
What should you track?
Keep a simple sheet with one row per creative:
- Concept and variation labels, so you can see patterns by idea.
- Spend, purchases and cost per purchase.
- Click-through rate and frequency.
- Status: testing, winner, paused or retired.
After a few cycles, patterns appear. Maybe lifestyle scenes beat studio for your brand, or close-ups beat full length. That knowledge is the real return on testing.
How do you make that many variations?
Twenty versions of a winning concept would normally mean a shoot plus days of design. WhatRunsIndia makes on-model photos and finished ads from your existing Shopify product photo, in all the main sizes, so building a variation round takes an afternoon. Every image is checked against your product and approved by you before use.
FAQ
How many ads should I put in one Meta ad set?
On a small budget, around three to six at a time works better than dozens, because each ad then gets enough delivery to judge.
How long should each creative testing round run?
About a week is a practical default, or until each ad has a meaningful number of purchases. Very short tests mostly measure luck.
What metric should decide a creative test?
Cost per purchase, or cost per your main conversion. Click-through rate is useful early but can favour ads that attract browsers rather than buyers.