What is Spend cap?

A spend cap is a ceiling on total spend: for an account, a campaign, a period, or the life of a campaign. When the total reaches the figure, delivery stops.

It is worth separating from a budget, because the two are often spoken of together and they do opposite work.

A budget is an instruction to spend. A daily budget tells the platform roughly how much to use each day, and the platform paces against it, deliberately spending above on days with better opportunity and below on weaker ones so the period total lands near the target. It is guidance about how fast to go.

A spend cap is a refusal. It does not shape pacing, it does not distribute anything, and it has no opinion about what the money achieves. It stops.

The failures a spend cap catches are financial rather than performance-related, and that is the whole category.

A misconfigured budget. A daily figure typed with an extra digit will spend ten times what was intended, and it will do so at a perfectly respectable cost per result. Every cost control in the account will report that things are fine.

A duplicated or forgotten campaign. A campaign cloned for a test and left running, or a seasonal campaign that outlives its season, spends steadily and reports nothing unusual.

An automated strategy with a loose target. A strategy told to maximise volume within a budget will use the budget. If the budget was wrong, the strategy will faithfully spend it.

A timing or currency mistake. An account configured in the wrong currency, or a schedule that runs on a different timezone than intended, produces spend nobody planned at prices that look normal.

The common thread is that none of these is visible to a cost-per-result control. They are all things going wrong at an acceptable price, which is precisely the blind spot a financial ceiling covers.

High. Higher than feels natural.

The purpose is to catch a mistake, and a mistake is by definition outside normal operation. A cap set close to the expected spend will bind during a legitimately busy week — a promotion, a seasonal peak, a successful test being scaled — and stop delivery for no reason at all, at the worst possible moment.

A useful way to pick the number: take the largest amount the account could spend in the period if everything went well, and set the cap meaningfully above it. Then treat the cap being reached as an incident to investigate rather than as a normal occurrence.

An account whose spend cap binds regularly has a cap set as a budget, and should replace it with a budget.

Three controls with three different promises, and only one of them is financial.

Bid cap — the price of any single auction. Binds every auction.

Cost cap — the average price of a result. Binds on aggregate.

Spend cap — the total. Binds on the sum, and says nothing at all about price.

A campaign can sit comfortably inside a spend cap while buying results at a price the business cannot afford, and it can sit comfortably inside a cost cap while spending far more in total than anybody intended. The two controls do not substitute for each other and an account that cares about both needs both.

The distinction matters in practice.

An account-level cap is the backstop. It catches everything, including the campaign nobody remembered, and it is the one worth having even in a well-run account.

A campaign-level or lifetime cap is narrower and more useful for a specific commitment: a promotion with an agreed total, a client budget that must not be exceeded, a test whose cost was sized in advance. Here the cap is closer to a contract term than to a safety rail, and setting it at the agreed figure is correct rather than conservative.

Delivery stops, abruptly and without regard to performance.

That is the right behaviour for a safety rail and it has consequences worth anticipating. Automated strategies that were mid-learning lose their momentum. Campaigns that stop and restart re-enter a period of unstable delivery. Anything time-sensitive — a promotion, an event — simply stops running.

Because of that, a cap that binds is worth being told about rather than discovering later. Where the platform can alert on approaching the ceiling, that alert is worth more than the cap itself, since it converts an abrupt stop into a decision somebody makes.

Set it, set it high, and review it when the account's normal scale changes.

The failure mode is not a cap that is too high — a high cap is inert and harmless, and still catches the extra-digit mistake. It is a cap set to the scale of an account two years ago, quietly throttling a campaign that has since grown into it, with nobody thinking to look at a setting that was correct when it was made.

A spend cap binding looks, from inside the campaign, like several other things.

Delivery ceases. Impression share collapses. The lost-to-budget column fills. Cost per result, computed over the period before the stop, looks normal or even good. Nothing in the campaign's own settings has changed.

An account that has forgotten the cap exists will look for the cause in bidding, in targeting, in creative approval status and in the auction, and will find nothing, because the constraint is one level up.

The check is quick and belongs early in any investigation of a campaign that stopped without explanation: look at the account's spend against any ceiling set on it, and look at the dates. A campaign that stopped at the same moment as every other campaign in the account was not stopped by anything inside itself.

Spend caps are the crudest control in the account and the only one that fails safe.

Every other control shapes what gets bought and depends on the surrounding configuration being sane. A target is only as good as the conversion it points at. A cost cap is only as good as the result it counts. A bid cap is only as good as the relationship between price and value.

A spend cap depends on nothing. It does not need tracking to work, it does not need a model, and it is not fooled by a mismeasured conversion. When everything else in the account is wrong at once — which is the situation it exists for — it is the only thing still holding.

That is an argument for having one even in an account that does not expect to need it, and for setting it high enough that its existence is never a reason for a campaign to underperform.