What is Daily budget?

A campaign set to A$200 a day spends A$247 on a Tuesday with strong demand and A$158 on the following Sunday. Neither day is an error: the platform is pacing to the average and using the opportunity where it exists. Over thirty days the expected total is about 200 × 30 = A$6,000, and it is the period figure rather than any single day that the instruction actually governs.

A daily budget instructs a campaign how much to spend per day on average. It is not a hard limit, and the most common surprise it produces is a day where spend came in above the number.

Platforms pace against the budget across a period rather than enforcing it each day. Where a day presents better opportunity — more auctions, cheaper inventory, higher predicted conversion rates — the system is permitted to spend above the daily figure, and it compensates on weaker days so the period total lands near the budget multiplied by the days elapsed.

Single-day overspend is normally capped at a modest multiple of the budget, so the variance is bounded rather than open-ended. The correct check is the period total. A campaign that spent 24% above budget on Tuesday and 21% below on Sunday is behaving exactly as instructed.

Inside a day, the platform decides when to spend. Two modes are common.

Standard pacing spreads delivery across the day, holding back so the budget is not exhausted by mid-morning. It produces more even delivery and captures opportunity across all hours.

Accelerated pacing, where it is still offered, spends as fast as the auction allows. It buys the earliest available inventory, which is not the best inventory, and exhausts the budget before the hours with better response arrive.

The default is almost always the right choice. Accelerated delivery is occasionally correct — a time-limited offer, an event with a fixed window — and is otherwise a way to convert a budget into whatever inventory happens to be available first.

Where the lost-to-budget column shows losses concentrated in a narrow band of hours, pacing rather than budget size is the thing to look at.

On a campaign running a manual bid, a budget change is arithmetic: more money buys more of the same.

On an automated strategy it is not. The bidder has been optimising against a spend rate, and changing that rate materially changes the problem it is solving. Large changes typically push the campaign back into a learning state, where predictions are being rebuilt, delivery is unstable, and the reported cost per outcome is unrepresentative of anything.

An account that responds to a bad week by cutting budget, then to the resulting low volume by raising it again, will keep a campaign permanently unsettled and never see its actual performance.

The practical discipline is to move budgets in steps of twenty or thirty percent rather than in multiples, to space changes by at least the conversion window, and to change one thing at a time. Where a large increase is genuinely needed, doing it in two or three steps over a fortnight costs a little time and avoids a week of noise.

There is a lower bound below which a daily budget cannot produce a usable campaign, and it is set by the price of the event being optimised toward.

A campaign told to optimise for a conversion that costs fifty dollars, funded at sixty dollars a day, can produce roughly one conversion a day. The variance on that is enormous, the bidder has almost nothing to learn from, and the reported cost per conversion will swing wildly from week to week with no underlying change.

The usual guidance is that a daily budget should comfortably cover several of the optimisation events. Where it cannot, there are two honest options: consolidate budget into fewer campaigns so that each has enough to work with, or optimise toward a shallower and more frequent event until volume supports the deeper one.

Spreading a small budget thinly across many campaigns, each below the threshold, is the most common structural mistake in small accounts and it guarantees that none of them ever learns anything.

Most platforms allow budget to be held at the campaign level or at the level below it, and the choice is about who decides the split.

Campaign-level budget lets the platform allocate across the ad sets or ad groups beneath it, moving money toward whatever is performing. This is usually more efficient and it occasionally produces an outcome nobody wanted — everything concentrated in one audience while the others starve, which is efficient in the short run and leaves the account with no diversity when that audience fatigues.

Budget held at the level below guarantees each segment gets funded. It protects a deliberate structure — separate prospecting and retargeting, separate geographies, separate product lines — at the cost of some efficiency, because money cannot flow to where it would work best.

Neither is correct in general. The question is whether the structure encodes a decision worth protecting.

A daily budget interacts with what the account is willing to pay in a way that is easy to misread.

A campaign that is not spending its budget is not necessarily under-funded. It may be unable to find auctions it expects to win at its target, in which case the constraint is the target rather than the money, and raising the budget changes nothing.

Conversely, a campaign exhausting its budget every day is telling you that at its current competitiveness there is more to buy. The lost-to-budget column sizes it, and the decision to fund it rests on whether the marginal conversion is still under the ceiling the contribution allows.

Reading spend, the budget-loss column and cost per conversion together answers which situation applies. Reading any one of them alone supports the wrong intervention about half the time.