Facebook Ads Budget: How Much Should You Spend? is the planning guide behind the Facebook Ads Budget Calculator on this page, and it starts from the answer most advertisers resist: there is no universal right budget, there is only the budget your conversion economics require. This article builds that number properly, from target CPA and Meta's learning-phase volume, checks it against the most your margin allows you to pay per customer, and then layers the testing and scaling decisions on top.
The floor: what the learning phase costs
Meta's delivery system optimizes each ad set against its conversion event, and it needs volume to do it, conventionally about 50 conversions per ad set per week to exit learning and stabilize. That requirement prices your floor directly: weekly budget per ad set equals target CPA times 50. At a 2,000 a week, about 3,000 monthly budget split across four ad sets funds none of them to learning volume, so all four deliver erratically and the post-mortem blames creative that was never actually tested. The calculator's first job is to force the consolidation question early, while it is still cheap.
The ceiling: what a conversion may cost you
The floor says what Meta needs; your margin says what you may pay. Maximum profitable CPA equals revenue per conversion times gross margin: 60 CPA and not a cent more before each sale loses money. The spend plan lives between the two numbers, and the calculator marks the zones: a target at or below 80% of the maximum leaves headroom for the variance auctions always deliver; between 80% and 100% the plan is profitable but fragile; above the maximum, no budget size fixes the plan, because scaling a negative-margin CPA manufactures losses faster. Businesses with honest repeat purchase can stretch the ceiling by counting verified lifetime value in the revenue figure, but the stretch must be earned with retention data, not asserted to make a campaign plan pass.
The worked plan
Target CPA 120, margin 50%, CPM 4,000 a week, **60, with the 20 of headroom per conversion. This is a coherent plan: the budget funds learning, the target respects margin, and the impression volume confirms the audience must be large enough to absorb a million monthly impressions without frequency running away (the Ad Frequency and Reach Calculator prices that check). Change one input, a $70 target CPA, and the same plan fails its ceiling test at any budget: the article you are reading exists so that this discovery happens in a spreadsheet rather than in a month of invoices.
Testing budgets, separated from scaling budgets
New offers and new creative need a test budget with its own rules: size it to buy enough conversions for a decision (a common standard is spending two to three times the target CPA per creative concept before judging it, accepting that a test at sub-learning volume is directional rather than definitive), ring-fence it from the proven campaign's budget so experiments cannot starve what works, and set the kill criteria in advance (spend threshold, CPA ceiling, minimum impressions per variant) because tests without pre-committed stop rules run until they accidentally find a good day and declare victory. Once a winner clears the test at target economics, it graduates into the learning-funded structure above, and scaling proceeds in controlled increments, commonly 20 to 30% budget increases every few days, because abrupt multiplications throw the ad set back into learning and reset the stability you paid for.
What the budget cannot fix
A sufficient budget buys learning, reach, and statistical clarity; it does not buy a working offer. If the ceiling test fails, the levers are price, margin, conversion rate, and order value, in the business rather than in Ads Manager. If delivery fails at adequate budget and healthy frequency, the levers are creative and audience. The budget question, answered honestly with the calculator, mostly serves to separate those two situations before money is spent confusing them, and to give every stakeholder the same three numbers: the floor, the ceiling, and the plan between them.
Frequently Asked Questions
Is there a minimum budget Meta itself requires?
Only a nominal technical minimum (around a dollar a day for impression-based buying, more for conversion objectives in practice). The meaningful minimum is economic, not platform-set: the learning-phase floor your CPA implies, which the calculator computes and which is usually far above the technical floor.
Should I start at the full calculated budget?
If the plan passes the ceiling test and the offer is proven, yes: underfunding a sound plan buys noisy data and slow learning. If the offer is unproven, start with a ring-fenced test budget sized by the test rules above, and promote winners into the full structure. Starting big on an unproven offer is not confidence; it is skipping the experiment and keeping the risk.
How often should the budget be recalculated?
Whenever an input moves: CPA drift over a month, margin changes from pricing or fulfilment costs, a new ad set added to the structure, or seasonal CPM shifts (the fourth quarter routinely reprices impressions). The plan is a function of its inputs, and stale inputs produce confidently wrong budgets.