Meta's targeting got good enough that audience tinkering stopped being where the gains are. What moves cost per lead now is how many genuinely different ideas you put in front of the algorithm, and how quickly you stop funding the ones that lose.
The trap is testing without structure: three variations of the same idea with a different background colour, run for two weeks, producing a result nobody can act on.
Test angles, not variations
An angle is a different reason someone would buy. A variation is the same reason wearing a different outfit. Testing five variations of one angle tells you almost nothing; testing five angles tells you what your market actually responds to.
Five angles that work as a starting set:
- Problem — name the pain in the first two seconds, before mentioning yourself
- Proof — a result, a before and after, a customer saying it worked
- Objection — address the reason they have not bought yet, directly
- Offer — lead with the deal, the guarantee or the risk reversal
- Comparison — you versus the obvious alternative, including doing nothing
Build one strong creative for each. Same offer, same landing page, same audience. The only variable is the angle.
Structure the test so the answer is readable
Run the five in one campaign, one ad set, letting Meta distribute. Purists will argue for isolated ad sets and even budget splits. In practice, at typical small-business budgets, splitting spend five ways produces five underpowered ad sets that all stay stuck in learning.
Set a total budget that can realistically produce around fifty conversions across the test. If that number is not achievable in two weeks at your cost per lead, test three angles instead of five.
Know what you are measuring, and in what order
Read the metrics in this sequence, because each one diagnoses a different failure:
- Hook rate (3-second views over impressions) — is the opening stopping anyone?
- Hold rate (thruplays over 3-second views) — does the middle keep them?
- Click-through rate — does the ad make them want to act?
- Cost per qualified lead — the only one that decides anything
An ad with a great hook rate and a terrible cost per lead is attracting the wrong people well. That is a different problem from an ad nobody watches, and it needs a different fix.
Call it, then iterate on the winner
When one angle is clearly ahead — not marginally, clearly — that becomes your control. The next week's work is variations within that angle: different hooks, different formats, different presenters, static versus video. Now variation testing is worth doing, because you are optimising something you know works.
Keep one slot open every week for a genuinely new angle. Winning creative fatigues, sometimes suddenly, and you want the next one already in testing when it does.
The cadence that makes it sustainable
Weekly, not monthly. Monday: review last week, kill the losers. Tuesday to Thursday: produce next week's concepts. Friday: launch. It sounds like a lot until you realise the alternative is running the same three ads for a quarter and wondering why costs keep climbing.
The accounts that win on Meta are almost never the ones with the cleverest targeting. They are the ones that shipped forty pieces of creative while their competitor shipped four.