CBO vs ABO: which is better in 2026?
Campaign budget optimization (CBO) lets Meta move budget between ad sets; ad set budgets (ABO) pin spend where you put it. In 2026 the honest answer is that consolidation made the debate mostly obsolete: an account running one to three ad sets per campaign, the structure the post-Andromeda platform rewards, barely has enough ad sets for the two options to differ. Where the choice survives, CBO is the default and ABO is a scalpel.
What actually changed underneath the debate?
The CBO-versus-ABO argument came from an era of ten-ad-set campaigns, one per audience hypothesis. Andromeda, Meta's rebuilt retrieval engine (announced December 2024, global by October 2025), moved audience selection into the auction itself, reading creative signals rather than your ad set boundaries. Fragmented structures now mostly fragment learning: each ad set needs enough weekly conversions to calibrate, and published panels showed consolidated accounts gaining roughly 8 to 10% just from restructuring.
So the modern question is not "who controls budget between my twelve ad sets" but "why do I have twelve ad sets." The furniture account in our restructure write-up went from 31 ad sets to 3, and its CPA stabilized in 19 days.
When is CBO the right call?
Almost always, once the structure is consolidated. With one to three ad sets feeding one objective, letting delivery shift budget toward cheaper predicted results is exactly the automation you want; overriding it hourly with manual reallocations re-adds the noise you removed. CBO also pairs naturally with Advantage+ sales and leads campaigns, which is where around 80% of the budget sits in accounts we run (the Advantage+ answer covers that split).
When does ABO still earn its keep?
- Guaranteed spend floors. A franchise location or product line that must receive budget regardless of auction efficiency.
- Clean experiments. A creative or landing-page test that needs an isolated, fixed budget to read.
- Deliberate market entry. Forcing spend into a new geography whose early CPAs will look worse than the incumbent markets.
Notice the pattern: ABO survives where a business constraint outranks auction efficiency. It is a governance tool now, not a performance tool.
What should you do with an old CBO/ABO account?
Stop optimizing the taxonomy and collapse it. The sequence we run: verify the event pipeline first, because consolidation aims delivery harder at whatever your events say (measurement service), then merge to one to three ad sets with CBO on, then raise creative volume to 10 to 20 live ads. Expect a rocky first two weeks while learning resets; our accounts median 24 days to a stable CPA. Budget changes after that move in steps of roughly 20% so the calibration holds.
What does the learning phase have to do with it?
Everything; it is the mechanism under the whole debate. Meta's delivery system needs roughly 50 optimization events per ad set within a rolling 7-day window to exit its learning phase, and an ad set that never gets there keeps paying the calibration tax indefinitely. Fragmented ABO structures fail this arithmetic by design: a $15,000 budget across ten ad sets at a $40 CPA produces about 37 weekly conversions total, spread so thin that no single ad set ever stabilizes.
The same mechanism explains the migration rules below: budget changes above roughly 20%, targeting edits, and bid-strategy changes all reset learning. CBO shifting budget between ad sets does not.
How do you migrate without wrecking a working account?
- Verify events first. Consolidation aims delivery harder at whatever your pipeline reports; fix duplicates and match quality before concentrating budget on them.
- Move in one restructure, not ten tweaks. Each structural edit resets learning; a single planned migration pays the calibration tax once.
- Hold spend flat for the first two weeks. Expect the CPA wobble and say so in advance; our accounts median 24 days to stability.
- Scale in steps of about 20% afterward, so the learning you paid for survives the raise.
The one-sentence verdict
Consolidate first; then CBO by default, ABO only where a business rule must outrank the auction, and neither as a substitute for creative volume and clean events, which decide far more of your outcome than budget plumbing ever will.
Short versions, for forwarding
Should I use CBO or ABO for testing creatives?
Neither, as a separate structure. Test inside the main consolidated campaign where possible: Meta's delivery gives new creatives an exploration budget, and judging them under real delivery beats a synthetic ABO test cell. We reserve ad-set-level budgets for the rare test needing a guaranteed, isolated spend floor.
Does CBO starve smaller ad sets?
It concentrates budget where predicted results are cheapest, which looks like starvation when an ad set covers a small audience. Sometimes that concentration is correct. When the "starved" segment matters commercially, like a specific metro for a franchise, give it its own campaign with its own budget rather than fighting CBO.
How many ad sets should a campaign have in 2026?
One to three. Post-Andromeda guidance and published panels agree: consolidated accounts with one to three ad sets and 10 to 20 creatives per ad set saw roughly 8 to 10% performance gains versus fragmented structures, mostly by clearing the learning phase faster and letting delivery aggregate signal.
If your account still has a CBO-vs-ABO folder taxonomy from 2021, the audit will pay for itself.
Book a working session