CBO vs ABO: which one to use and when

CBO and ABO describe where the budget lives: in the campaign or in each ad set. The choice looks technical and ends up defining how much real control you have over where the money goes.

The difference in one sentence

With ABO, you decide that ad set A spends fifty and ad set B spends fifty. The platform complies.

With CBO, you decide the campaign spends a hundred and the platform decides how much each ad set gets, correcting the split every day based on what's working.

Everything else — the entire debate — follows from that.

Why CBO usually wins when there's volume

The platform reallocates budget several times a day with information no person has: who's available in the auction right now, at what price, and with what conversion probability.

When there are enough results for that decision to be grounded, CBO allocates better. A media buyer reviewing the account once a day makes at most one reallocation decision per day; the platform makes dozens.

The argument for CBO isn't that it's "smarter" in the abstract — it's that it acts more often, with better information about the moment.

Why ABO survives

There are three situations where CBO disappoints, and none of them is about the algorithm being bad.

Low volume. With few weekly results, the platform has no evidence to distinguish a good ad set from a lucky one. It splits with noise, concentrates spend on whichever started better by chance, and the other never gets tested.

Needing guaranteed delivery. If a strategic audience has to get spend — a remarketing segment, a new region — CBO guarantees nothing. It can leave it near zero if another ad set performs better in the short run.

Testing. To compare two audiences or two offers you need comparable spend. CBO makes it uneven by design.

The mistake that makes CBO "not work"

The most frequent pattern: someone builds a CBO campaign with six ad sets, looks after three days, sees five barely spent, and concludes CBO is broken.

It isn't broken — it's doing exactly what it was asked. With six ad sets and a budget sized for modest volume, the platform concentrates where it sees signal and abandons the rest. The structure asked for a choice, and the choice got made fast, on thin evidence.

The fix isn't going back to ABO: it's reducing the number of ad sets. CBO with two or three ad sets, each with enough volume, behaves far more reasonably than CBO with eight.

A decision rule that works

Count the optimized results the campaign produces per week.

Under fifty in total, use ABO. There's no volume for any automatic allocation to be grounded, and manual control is at least predictable.

Comfortably above fifty, with room for each ad set to accumulate its own signal, use CBO with few ad sets.

If you're testing something — new audience, new offer, new format — use ABO for the duration of the test, whatever the volume. The test needs an even split; the operation needs an efficient one. Those aren't the same goal.

Spend limits per ad set

Meta lets you set minimums and maximums per ad set inside a CBO campaign. It sounds like the perfect fix for guaranteed delivery, and it should be used carefully.

Every constraint you add takes freedom away from the algorithm, and past a certain point CBO with many limits performs worse than plain ABO — you pay the cost of automation without getting the benefit.

A minimum on one strategic ad set is reasonable. Minimums and maximums on every ad set is ABO written in a complicated way.

What changes in the report

Worth anticipating one operational annoyance: with CBO, budget per ad set varies day to day, so a report comparing ad sets needs to look at cost per result, not absolute spend.

An ad set that spent three times more than another isn't "the best" — it's the one the platform picked. The performance question is still the same: what did each result cost.

This detail confuses a lot of people in meetings, especially when whoever reads the report is used to ABO and assumes the split was a team decision.

What a well-built CBO account looks like

In practice, accounts that get value out of CBO share a shape: few campaigns, few ad sets per campaign, and many creatives inside each ad set.

The logic is to let the platform decide where it has information — which creative to show to whom — and keep the structural decisions on the human side: which audiences are worth having, what offer to make, how much total budget the channel gets.

An account fragmented into twenty ad sets is asking the algorithm to make twenty decisions with a twentieth of the evidence each. It's the most common way to get bad results from a tool that works.

The remarketing case

Remarketing deserves its own note, because it's where CBO disappoints most predictably.

Remarketing audiences are small and usually have better cost per result than prospecting. In a mixed CBO campaign, the platform tends to pour budget there — exhausting the audience in days and pushing frequency to an absurd level.

Separating remarketing into its own campaign, with its own budget and a deliberate ceiling, almost always beats letting it compete with prospecting inside the same pot. Not because CBO fails, but because those two audiences have different jobs and the algorithm optimizes a single number.

Why the debate keeps coming back

One reason this argument never settles: both sides are describing real experiences from different accounts.

The agency running fifteen small local clients genuinely gets better results with ABO, because none of those accounts produces the volume CBO needs. The team running one large ecommerce account genuinely gets better results with CBO, for the same reason in reverse.

Neither is wrong about their own account, and both generalize. When you read a strong opinion about CBO, the useful question is how much weekly volume the person's accounts produce — that single number explains most of the disagreement.

Switching between them

Moving from ABO to CBO or back resets campaign learning. That's not a reason to avoid it, but it is a reason not to do it in month-end week or in the middle of a seasonal push.

The sensible moment is the start of a cycle: new creatives going live, new month, new structure. The reset was going to happen anyway.

And give it time before judging: the first days after the switch say nothing about whether the decision was good.

To understand why structural changes are expensive, see the learning phase. If remarketing came up, continue with cold, warm and hot audiences. And for the weekly read without exporting, see the paid traffic report.