How *campaign budget* puts spend where it wins

What centralised budgeting does#

Advantage campaign budget, formerly CBO, holds one budget at campaign level and shifts it continuously toward the ad sets and ads with the best predicted marginal return. Winners pull spend within hours while laggards fade without manual reallocation. Meta details the behaviour in its campaign budget guidance, and the Meta Ads hub places it in the family.

Centralisation compounds learning: fewer budget fragments mean each active ad set keeps a clearer shot at 50 weekly events. Structure gets simpler while creative range matters more.

How allocation really behaves#

Allocation follows predicted marginal return, not lifetime loyalty: this morning winner can lose budget by afternoon when its next conversion looks dearer than a rival. New ads get exploration spend to prove themselves, then either earn their share or fade. Expect movement, and judge on weekly campaign totals rather than hourly ad set lines.

Minimums still apply: each active ad set needs enough to stay meaningful, so cap the ad set count by budget. Learning Limited flags tell you when the pool is spread too thin.

When centralised wins biggest#

Centralised wins with consolidated structures and real creative range: two to four ad sets, several concepts, one commercial objective. It also wins for evergreen prospecting where audiences overlap, because the budget routes around overlap instead of funding both sides of it. Consolidation sets the ad set count, and Advantage+ sales pairs naturally with it.

Keep it for scale phases where speed of reallocation beats human review cycles. Weekly planning plus daily monitoring is the rhythm, not hourly fiddling.

Minimums and guardrails#

Size the campaign so each ad set could plausibly clear 50 events if it won a fair share: divide weekly budget by ad set count and check against target CPA. Use spend limits per ad set sparingly, because heavy minimums recreate fixed budgets with extra steps and starve the reallocation you bought. Meta troubleshooting in the learning phase guide confirms the volume maths.

Keep test cells outside the pool in clean A/B builds with ad set budgets, so experiments never fight the allocator for spend. Document every limit with its expiry date.

Monitoring without meddling#

Review campaign totals weekly, ad set share trends and creative fatigue in that order: totals judge the budget, shares reveal winners, fatigue explains decay. Resist daily reallocations, because the allocator needs stable history to route well. Significant edits restart learning per edit rules, so batch structural changes.

Step in where control matters: strict geos, margin guards and test integrity. Ask our team for a budget structure review when shares swing without reason.

Creative testing beside the pool#

The pool rewards range, so run a separate fixed-budget test cell that feeds winners into it: three new concepts against one control, equal funding, full seven-day reads, then graduate victors into the centralised campaign without editing them. This keeps exploration spend quarantined while the pool compounds proven winners. The cadence in 322 testing fits small teams, and diversification defines what counts as a genuinely new concept.

Never test inside the pool itself, because the allocator starves unproven ads before they gather evidence and the read corrupts. Log every graduation with its test numbers so the pool history stays legible. Clean A/B rules govern the handoff, and our team audits handoffs quarterly. Patience at the handoff beats enthusiasm, because premature graduates waste pool spend proving what the test cell would have shown cheaper.

Frequently asked questions#