Should I run separate main and test campaigns on Meta ads?
Short answer
Yes, once you have a winner. I always run two campaigns: a main campaign with the proven ad that I never touch, and a test campaign for new ideas. Put 75% to 90% of spend in main, about $5 a day in test. When a test ad wins, move it to main. Beginners can skip testing until the first ad has done its work.
We normally have two campaigns no matter what. Once a campaign is working, you don't want to mess with it. So we have main and test.
"Don't touch the one that works; test in the second campaign."
How does the structure work?
| Campaign | Budget | Job |
|---|---|---|
| Main | $20 to $30 a day, 75% to 90% of spend | Runs the proven ad. Left alone. |
| Test | About $5 a day, 10% to 25% of spend | Random ideas, new creative, hyper-targeted versions. |
If something in the test is working really well, it graduates to main.
Why not test inside the main campaign?
Because adding anything to a working campaign starts the algorithm over. The test campaign's job is to protect the main one while you try new creative.
You also need the test lane for a quieter reason: the main campaign's cost drifts up over time as the reachable audience gets used up. You want fresh creative already proven and waiting when that happens.
Do beginners need a test campaign?
Not really. When I coach someone just starting, I tell them to run the winner and not even worry about testing.
"Anybody starting out, which is almost everybody in local newsletters, you don't need to try anything."
I only test more now because we run 15-plus brands. I've tested a lot, and nothing has beaten the ad I already have. When you want something hyper-specific, like an ad calling out one suburb by name, that's when you throw it in a test campaign.
What if my budget is small?
Consolidate. On one small-budget account my media buyer and I collapsed several campaigns into one campaign with two ad sets, because splitting a small budget across multiple campaigns means none of them gets enough delivery data. He also staggers new creatives about ten days apart, since changing too often hurts performance.
Can I test faster on someone else's money?
On client brands where the client doesn't care about squeezing every dollar, I tell my team that doing more is better: more creatives, more text, more locations. Volume of testing matters more than any one test working perfectly. Only proven variants come back into my own brands. My process for what to test is in how to test new Meta ad creative.
Answered by TJ Larkin. From publisher setup calls and media-buyer working sessions in 2025 and 2026. Last updated .