Why Is One Ad Set Getting All the Budget? (CBO Explained)

When you set your budget at the campaign level, you are using what Meta now calls Advantage+ campaign budget, the feature long known as CBO (campaign budget optimization). You give Meta one pot of money and permission to move it between ad sets however it likes. Meta's developer documentation describes the mechanic plainly: the system "automatically and continuously finds the best available opportunities for results across your ad sets and distributes your campaign budget in real time to get those results." In practice, that means Meta is constantly predicting which ad set will produce the next conversion, lead, or click at the lowest cost, and steering the next ringgit there. If one ad set has a stronger creative, a more responsive audience, or simply got off to a faster start, the prediction engine favours it, it gets more delivery, it generates more data, and the prediction gets more confident. The loop compounds. A 70/20/10 split, or even 95/5/0, is a normal outcome, not a bug. If you wanted equal spend per ad set, campaign-level budgeting was never designed to give you that.

Often the concentration is correct and you should leave it alone. The favoured ad set is genuinely your best performer, and forcing money into weaker ad sets would just raise your blended cost per result. Before intervening, check the numbers: if the dominant ad set's cost per result is at or below target, the algorithm is doing its job. It becomes a problem in a few specific situations. First, testing: if you built three ad sets to compare audiences or creatives, a 95/5/0 split means two of them never received enough delivery to produce a readable result, so you learned nothing. Second, early-signal bias: Meta's predictions early in a campaign rest on thin data, and an ad set that got lucky in its first hours can lock in the budget before a genuinely better ad set has a chance. Third, misaligned wins: the favoured ad set may be producing cheap results that matter less to you, for example inexpensive leads from an audience that rarely buys, while the starved ad set targets your actual buyers. Fourth, audience overlap: when ad sets chase the same people, one tends to win those auctions repeatedly and absorb the budget, which looks like performance but is partly structural. The tell in every problem case is the same: the starved ad sets never accumulated enough results for you to judge them fairly.

Meta gives you a middle ground before you abandon campaign budgets entirely: ad set spend limits. You can set a minimum spend target so an ad set keeps getting delivery, or a spend cap so a dominant ad set cannot swallow everything. Two caveats, both straight from Meta's documentation. Minimums are a best effort, not a guarantee; Meta states the target "does not guarantee you spend this amount." And every limit you add constrains the optimization you turned on campaign budgeting to get, so use them sparingly, on the one or two ad sets you specifically need protected, rather than on everything. If you need guaranteed, even spend, use ad set budgets (ABO) instead. Each ad set gets its own budget and Meta cannot move money between them. This is the standard structure for creative and audience testing, where equal exposure matters more than efficiency. A common working pattern is to test in a small ABO campaign, then move proven winners into an Advantage+ campaign budget campaign and let Meta concentrate spend among ad sets that have all earned it. Finally, consider whether the structure itself is the issue. Merge overlapping audiences so ad sets stop competing for the same people, and cut ad sets you already know are weak. And before forcing budget into a starved ad set, ask whether its creative deserves the rescue: researching what similar brands run, for example through AdPlay.ai's archive of Malaysian Meta ads, is often faster than paying Meta to test an angle the market has already ignored.