The NEW BEST Facebook Ads Campaign Structure for 2026

Summary

Charlie presents a Meta campaign structure designed to reduce noisy testing and scale based on business results—not just ad-level ROAS or CPA. His central argument is that advertisers often stall by repeatedly testing and replacing ads before establishing whether a test improved campaign profitability.

Campaign Structure and Testing Playbook

Use a CBO campaign with a simple control-and-tests structure:

• Control: A small set of verified, relatively stable ads. These are benchmarks, not permanent winners.
• Tests: In the advanced setup, use two test ad groups, each containing one 3-2-2 Flex ad (rendered as “322” in the subtitles). Each test should outperform the weakest ad in the control—not necessarily find a “unicorn.”

The presenter advises against using interest groups or lookalikes in this approach.

Three Steps for Testing

1. Check delivery: Did the test ad spend its budget? If not, treat it as uncompetitive and restart with another test.
2. Evaluate the result: If it spent, did campaign profit increase? If yes, the test succeeded; if not, it failed. Individual ROAS or CPA does not determine the winner.
3. Scale successful tests: After a successful test, pause testing temporarily and increase the budget while the campaign continues to improve. The creative test becomes a stress test to find out how much more budget it can support.

End the test only when scaling is no longer possible, then review the ads for replacement candidates.

To identify a weak ad, compare its gross profit per transaction (GPT) with the campaign average. Prioritize high-spending ads with below-average GPT and above-average CPA, then test a replacement against that weak link.

If a test fails, turn it off without changing the control. The speaker describes disabling a failed test as a non-destructive edit that preserves the existing setup and momentum.

Decision Principle and Metrics

The main success question is: Did campaign revenue or profit increase?

The speaker defines gross profit as total revenue minus total advertising costs and emphasizes total profit volume as the test outcome. ROAS and CPA are diagnostic indicators, not sufficient proof that a test improved the campaign.

At the ad level, use GPT relative to the campaign average, alongside CPA, to help identify ads that may need replacing. The speaker provides no target ROAS, CPA, or GPT values; decisions are based on campaign results and comparisons with the campaign average.

Structure by Conversion Volume

The appropriate setup depends more on conversion volume than on budget:

• Level 1 — Simple: One ad group containing both control and test ads. Stay here until two groups can consistently exit the learning stage.
• Level 2: One control ad group and one test ad group. Stay here until conversion volume supports three groups.
• Level 3 — Advanced: One control ad group and two test ad groups. The presenter says he would keep this structure even at $1 million in monthly spend rather than add more complexity.

For catalog ads, the exact Flex-ad format may not be available, but the same control-and-test logic and profit-based evaluation still apply.

If Flex ads are missing, the speaker recommends enabling dynamic creatives at the ad-group level; the Flex option should then appear at the ad level.

Examples and Claims

The speaker contrasts two campaigns with the same product, audience, ads, and starting daily budget but different campaign structures. He says the account using the recommended “Andromeda One” structure performed substantially better over time, though he provides no specific results.

To illustrate why budget alone can be misleading, he cites campaigns spending $1,000 per day with 30 sales per week and $200 per day with more than 65 sales per week. He argues that the lower-conversion campaign needs a simpler setup despite its higher spend.

He frames the broader operating mindset as moving from “hunter” marketing—reacting to daily performance—to “farmer” marketing: building a system and making fewer disruptive changes so results can compound.

Presenter/source: Charlie, identified with the Professor Charley T channel. The examples and recommendations are based on the presenter’s account experience and Meta Ads campaign setup.