Everyone Tells You to Use CBO. Nobody Says What It Decides.
There is one toggle in Meta Ads Manager that decides who allocates your money, and most owners have never knowingly touched it.
When you build a campaign, Meta asks where the budget lives. Put it at the campaign level and Meta decides how to split it across your ad sets. Put it at the ad set level and you decide, and Meta spends what you told it to spend whether that was a good idea or not.
The industry shorthand is CBO for the first, campaign budget optimization, and ABO for the second, ad set budget optimization. The names are forgettable and the choice is not. It determines whether your budget follows performance automatically or sits exactly where you last left it.
Search for advice on this and you will find a lot of people telling you to use CBO. Some of them run large ecommerce accounts. Very few of them are running your budget.
What Each One Actually Does to Your Money
Strip out the acronyms and it is simple.
With the budget at the campaign level, you give Meta one number for the whole campaign. Say a hundred dollars a day. Meta looks at your ad sets, forms a view about which is most likely to produce the result you asked for, and moves the money there. You might have ten ad sets and find that Meta spent almost all of it on two of them and nothing on the rest. That is not a malfunction. That is the feature.
With the budget at the ad set level, you give each ad set its own number. Four ad sets at twenty-five dollars each. Every one of them spends its twenty-five whether it is working or not, until you go in and change it.
So the real question is not which acronym performs better. It is a question about you:
For most owners the honest answer is yes, and that is the case for CBO. But it is not the whole answer, because it assumes there is enough money moving for reallocation to be the thing that matters.
The Way Ad Set Budgets Actually Fail
Sam Piliero's agency runs close to a hundred Meta accounts. He published a walkthrough of an account they took over that had been running entirely on ad set budgets, and the failure in it is worth looking at closely because it is not the failure people expect.
Across a four-month period the account spent about thirty-six thousand dollars. One ad set, returning 1.28 on ad spend, had absorbed five thousand of it. It was one of the worst performers in the account and it was in the top three for spend.
Nobody chose that. Somebody set that ad set's budget months earlier, when it looked promising, and then never went back. Meanwhile other ad sets in the same campaign were returning close to 2.0 and were capped at whatever number they had been assigned on the day they launched.
That is ABO's actual failure mode, and notice that it is not a technical one. The system did exactly what it was told. The problem is that keeping ad set budgets honest requires somebody to sit down every week, compare performance across ad sets, and move money by hand. In a business where you are also doing the work, answering the phone and paying the bills, that review is the first thing to get skipped.
The Way Campaign Budgets Fail, Which Nobody Mentions
Now the other side, which the advice tends to leave out because most of the people giving it are not spending thirty dollars a day.
Give Meta one budget across several ad sets and it concentrates. On a large budget that is exactly what you want. On a small one it is a problem, because concentration and elimination look identical when there is not much money to go around.
Say you are spending thirty dollars a day across four ad sets. Meta forms an early view, sends most of it to one, and the other three receive a few dollars each. Those three will never accumulate enough results to prove anything. At the end of a fortnight you have one ad set with data and three with noise, and you have learned almost nothing about the three.
There is a second, quieter version of this. Meta's early view is formed fast, and it is formed on how people respond as much as on what they buy. An audience that would have worked can lose the allocation in the first day and never get a second look, because there was never enough budget for a second look to happen.
None of that makes CBO wrong. It makes CBO a thing you graduate into.
So It Is a Threshold, Not a Philosophy
Here is the framing that resolves the argument, and it comes from watching what practitioners actually do rather than what they recommend in the abstract.
Chase Chappell describes starting audiences in ad set budgets and migrating them to campaign budgets once an audience is crossing a couple of hundred dollars a day in spend. That is the whole answer in one sentence, and it is a threshold rather than a belief.
The reasoning holds up. Automatic reallocation is valuable in proportion to how much there is to reallocate and how many proven options it has to choose between. Early on you have neither. You do not know which audience or angle works, so there is no baseline for Meta to protect, and your budget is too small to fund several ad sets far enough to find out. What you need at that stage is a guarantee that each thing you are testing actually gets spent on. That guarantee is exactly what ad set budgets provide.
Later, the position inverts. You know what a good cost per result looks like. You have a proven ad set carrying the account. Now the valuable thing is not forcing money into every idea, it is making sure new ideas have to beat the proven one before they get funded. That only works if the new ideas are genuinely different rather than variations on the winner you already have, which Meta bundles together and funds as one. Campaign budgets create that competition automatically, and they do it every hour rather than whenever you next open the account.
The practical rule. Ask how much daily spend each ad set would get if the budget were divided evenly. If that number is comfortably above your target cost per result, campaign budgets can work, because every ad set can still produce results often enough to be judged. If it is below, use ad set budgets, or run fewer ad sets.
Lead Generation Changes the Answer, and So Does a High Price
Two exceptions matter enough to name, because both come from people arguing against their own convenience.
The first is lead generation. The team at Leadbase, who buy for lead-gen businesses, tested campaign budgets against ad set budgets in their own account and published the result: their campaign-budget test produced leads at roughly three times the cost of their ad-set-budget campaigns. They are upfront that the test was not perfectly matched. But their read is that campaign budgets suit ecommerce and many-audience accounts, and that lead gen behaves differently, partly because a lead is a much cheaper and noisier event than a purchase, so Meta's early confidence is built on a signal that correlates loosely with whether the lead was any good.
The second is high ticket, and it comes from Piliero, who otherwise puts about ninety percent of his spend into campaign budgets. His carve-out is that very high ticket businesses, where a sale is a thousand dollars or more, often need ad set budgets, because conversions arrive too slowly for the algorithm to learn before it has already decided.
Both exceptions share a cause. Campaign budgets are a bet that Meta will learn faster than you can review. When conversions are rare, slow or unreliable as a quality signal, that bet gets worse, and holding the budget yourself gets better.
What Switching Actually Costs
If you decide to move, know that this is not a cosmetic edit.
Changing where the budget sits is a significant change to a campaign, and delivery does not carry over untouched. Ad sets that were performing steadily can behave unpredictably for several days while the system settles into the new arrangement. If you make the change on a Thursday and panic on Saturday, you will turn off something that was about to be fine.
So do it deliberately. Move when the account is stable rather than while you are already firefighting, change nothing else in the same week, and give it a clear seven days before you judge anything. If the campaign carrying your revenue is working, the lower-risk route is to build the new structure alongside it and shift budget across gradually, rather than converting the thing paying your bills and hoping.
And if you are on ad set budgets and simply have not reviewed them in a while, do that before you conclude the structure is the problem. A weekly ten-minute pass moving money from the worst ad set to the best captures most of what campaign budgets would have done for you.
This Decision Also Decides Which Levers You Get
One consequence worth knowing, because it catches people out.
Several of Meta's controls only exist when the budget sits at the campaign level. Ad set spending limits are the important one: the maximum that stops a single ad set eating everything, and the minimum that forces spend into an ad set Meta is ignoring. Those only appear on campaign-budget campaigns, for the obvious reason that on ad set budgets you already control the number directly.
Which means the budget decision is upstream of how you test. We wrote separately about how to force a fair shot for a new ad, including how those spending limits behave and the traps in them. That article assumes campaign-level budgets throughout. If you are on ad set budgets, your version of the same problem is simpler and cruder: you give the new ad set its own budget and it spends it. And the reason a new ad struggles to get seen in the first place is Meta picking a favorite early, which happens either way.
What to Do on Monday
Open Ads Manager and look at your main campaign. If there is a budget figure at the campaign level, you are on campaign budgets. If the figures sit on each ad set instead, you are on ad set budgets. Most owners genuinely do not know which they are running, so start there.
Then do one piece of arithmetic. Divide your daily campaign spend by the number of live ad sets, and compare it to your target cost per result. If the answer is that each ad set could produce several results a day, campaign budgets are reasonable. If the answer is that an ad set might produce one result every three days, you are spread too thin for Meta to allocate sensibly, and the fix is fewer ad sets before it is a different budget setting.
If you are on ad set budgets, open them and ask when you last changed those numbers. If it was more than a month ago, you are not running ad set budgets, you are running a decision you made once.
And if you are early, still working out which audience and which angle, stay on ad set budgets a while longer. Guaranteed spend on the things you are testing is worth more right now than automatic reallocation between things you have not proven.
If you would rather someone watched this and moved the money for you, that is what our Meta ads management does all day. And if you would rather run it yourself, the arithmetic above is genuinely the whole decision. Either way, go and find out which one you are on.
