When *ad set budgets* beat centralised spend

What fixed budgets do well#

Ad set budgets guarantee spend where you promise it: each audience gets its funded share regardless of early results, so challengers survive long enough to prove themselves. That guarantee makes them the right tool for diagnosis, fair tests and strict control briefs. The Meta Ads hub places them beside centralised options, and Meta documents both in its campaign budget guidance.

Use them where the question matters more than the week revenue: which audience, which angle, which offer earns consolidation. Fixed budgets buy answers; centralised budgets buy scale.

When control beats centralised#

Control wins for head to head tests, tight geos and regulated briefs: A/B cells need equal funding, location splits need guaranteed coverage, and learning diagnostics need isolated variables. Small accounts with one or two ad sets also gain little from centralisation, because there is nothing to reallocate between.

Keep the fixed phase timeboxed: diagnose, decide, then consolidate winners into campaign budget for scale. Permanent testing mode is just fragmentation with better branding.

Designing fair tests#

Hold budget, audience size and creative count equal across cells, change one variable at a time, and run a full seven-day cycle before reading. Unequal budgets buy unequal learning, which flatters the funded cell and buries the challenger. Log hypothesis, spend and verdict per cell so the library compounds.

Size each cell to roughly 50 events a week, borrowing the maths from learning phase, and isolate tests from evergreen spend. Meta test guidance plus the learning phase guide confirm the volume needs.

Minimums and fragmentation guards#

Every fixed ad set must still fund 50 optimisation events a week or accept Learning Limited status: three starved ad sets teach less than one funded survivor. Merge lookalike slivers, stack interest themes per consolidation, and kill cells that spend 3 to 5 times target CPA with zero conversions.

Count active ad sets against weekly budget before every launch: budget divided by ad sets divided by CPA should clear 50. If it cannot, cut ad sets, not ambitions.

Graduating to centralised spend#

Promote proven audiences and angles into one consolidated campaign budget, pause the fixed originals rather than deleting them, and freeze edits for a full cycle so allocation history builds cleanly. Keep one fixed test cell running beside the scale engine for the next question.

Review the split quarterly with scaling plans: evergreen winners centralise, open questions stay fixed. Ask our team for a structure review when fixed and centralised fight over the same buyers.

Daily versus lifetime budgets#

Daily budgets pace evenly day by day, which suits evergreen tests and stable offers: delivery learns weekday patterns and spends through them. Lifetime budgets pace toward the end date, spending harder where conversions look likely, which suits launches, events and flighted pushes with a fixed close. Pick daily for diagnosis, lifetime for deadlines. First setup covers the mechanics.

Lifetime budgets need honest date ranges: too short a flight starves learning, too long one drifts without review. Size either type to 50 weekly events per ad set, watch frequency on small audiences, and read results on full cycles because attribution lag stretches short flights. Keep the budget type constant through a test, since switching types mid read restarts pacing history. Document the pacing choice with its reason, because future editors inherit the rhythm blind. Revisit pacing whenever the offer calendar changes, since launches and evergreen demand different rhythms.

Frequently asked questions#