Start with the purchase volume your budget can support and the differences between the ads you want to test. A batch of hook variations answers a narrower question than a batch built around different customer motivations. Count those separately, then check whether the budget can produce enough evidence to inform the next brief.

Start with what your budget can actually read

Campaign budget optimization sets a budget at campaign level and allocates it across ad sets. Delivery within each ad set can also be uneven. The question is how much evidence the available budget can produce for the decision you need to make, not how many ads the interface lets you load.

Use an illustrative budget of $300 a day and a $40 cost per purchase. If that cost holds, the campaign would produce about seven or eight purchases a day in total. Those purchases will not necessarily spread evenly across the ads. Loading fifteen variants does not give each one a useful test. Budget divided by a target CPA is a capacity estimate, not a forecast or a significance calculation.

If an ad gets no spend, first check delivery eligibility and account constraints. Clean statuses and spending siblings can indicate allocation preference. They do not establish that an untested concept cannot convert. Record that candidate as insufficiently tested before deciding whether it deserves a separate experiment.

Count concepts separately from hook variants

Separate concept exploration from iteration. Two executions of the same argument can be useful variants, but they do not answer the same question as two different arguments. Tag the customer problem, promise, mechanism and format before counting the bets. Otherwise three hooks can look like three concepts in the report.

A different format can change how an argument lands. It does not necessarily change the argument itself. A testimonial in video and a testimonial in an image may test execution while holding the underlying proposition constant. That is useful when execution is the question. A concept test needs a meaningful difference in what the buyer is being asked to believe.

Check that each candidate is finished, approved and eligible to deliver before including it in the test.

A real brief: three hooks, one concept

My ErgoSleeve DCT #2 brief, "Aesthetic Sarah," starts with a specific customer motivation: the laptop setup should belong with the rest of her desk and bag. The brief uses three openings into one shared body. One opens on an ugly stand disrupting a curated desk. Another starts with the surprise that an attractive object is functional. The third starts with the match between the setup, desk and bag.

The shared body carries the product explanation. That makes this one concept with three hook variants. It gives a creative analyst a narrower question to investigate: which opening helps this argument work? It does not test whether desk aesthetics beats every other customer motivation.

Decision	What this brief lets us examine
Concept	Desk identity and appearance as the entry point.
Variable	Three openings; the main body is shared.
Readout	Delivery and business outcomes for each variant, with the amount of evidence stated.
Next test	Preserve the body to investigate the opening, or use a separate concept to test a different motivation.

The first two rows describe the brief. The readout and next-test rows are the analysis it calls for, not results claimed from the document. A shared body improves the design of a comparison, but ordinary campaign delivery still does not randomize the audience or isolate the hook's causal effect. See the brief and resulting creative in my portfolio.

How I size a launch

Illustrative planning examples. The purchase counts assume the stated cost per purchase holds. The concept counts are starting points for judgment, not validated thresholds.

Read the table as a workload constraint. A wider batch needs more evidence and a team able to act on the findings. An ad count cannot guarantee a useful result inside a week. Decide what would count as a meaningful difference, the loss you can afford and when you will review the test before launch.

Plan the next batch while this one runs

A launch also needs a follow-up plan. Decide who will review the results, which questions the next batch should answer and how much finished creative the team can supply. Otherwise an initial batch of eight ads can turn into the same eight ads running for months.

A weekly review gives the team a regular place to discuss new briefs and results. It does not require replacing effective ads every week. If the team can finish three new ads, choose the three comparisons most likely to inform the next decision.

When an ad earns more spend while meeting the business target, evaluate scaling against margin, conversion lag and the return at the current spend level. A separate ad set can be a deliberate allocation choice, but duplication does not guarantee that the observed return will transfer. Define the reason for the move and its failure condition first.

If you truly need every variant read equally

When a decision requires a causal comparison between variants, use a properly designed randomized experiment. CBO and ad-set budgets control allocation; neither makes an ordinary group of ads a randomized test. Set the outcome, comparison, sample requirements and loss limit in advance. A small sample can remain inconclusive even when one reported ROAS is higher.

Meta's reference on how a campaign-level budget moves across ad sets: About Advantage+ campaign budget.