How much should a business spend on Meta ads?

Why nobody can hand you a number

A dentist, a candle brand, and an HVAC company can all run Meta ads profitably on wildly different budgets — because the budget isn't a property of the platform. It's a property of your economics. A business earning a few thousand dollars per customer can pay far more for a lead than one earning forty dollars per order, and a business whose customers come back monthly can pay more than one that sells to each customer once.

So when an article or a sales rep tells you "start with $X a day," they're answering without asking. The real question is: what can you afford to pay for a customer, and how many customers do you want? Everything else is derived from that.

The four numbers that set your budget

Before opening Ads Manager, pull these from your own records. None of them come from Meta.

1. Customer value

What is a new customer actually worth to you — not the first transaction, the relationship? A cafe's first ticket might be small, but if a meaningful share of first-timers become regulars, the honest value is a multiple of that ticket. Be conservative: use what your own history supports, not what you hope. If you have no idea how often customers return, that's worth fixing anyway — it's the same math that makes email and SMS retention worth running.

2. Gross margin

Ads are paid out of margin, not revenue. A retailer at thin product margins and a service business at wide margins can generate identical revenue per customer and have completely different room to advertise. Work out what you actually keep from a customer after the cost of delivering the product or service — before ad spend.

3. Tolerable acquisition cost

Decide what you're willing to pay, in margin dollars, to acquire one new customer. Some businesses want profit on the first transaction. Others — especially those with strong repeat behavior — will happily break even or go slightly negative on the first sale to own the relationship. Neither is wrong; what's wrong is not deciding, because then every ad result is judged by mood.

4. Conversion assumptions

Ads buy attention; your funnel converts it. If you're a lead business, roughly what share of leads become paying customers? If you're e-commerce, roughly what share of engaged visitors buy? Use real historical numbers wherever they exist, and mark every guess as a guess so you know which assumptions to check first when reality disagrees. Chain these together and you can translate "tolerable cost per customer" into "tolerable cost per lead" or "tolerable cost per purchase" — the number the ad account actually reports against.

A worked hypothetical

Imagine a home-services company where an average job is worth $2,000 in revenue at 50% gross margin — $1,000 of margin per customer. The owner decides they'll spend up to $300 of that margin to win a job. If, historically, one in four qualified leads becomes a booked job, then a lead is worth paying up to $75 for. That $75 is the line every campaign gets judged against — and the budget follows from how many jobs the business wants and can staff.

The point isn't these particular numbers — yours will differ. The point is that this business now knows what "working" means before spending a dollar, which puts it ahead of most advertisers on the platform.

The learning-phase reality: algorithms eat conversions

Here's where the budget question stops being pure arithmetic. Meta's delivery system learns who to show your ads to from the conversions your account generates. Feed it a steady stream of conversion events and it gets sharper; starve it and it stays in an expensive guessing mode indefinitely.

In practice, that sets a floor on a workable budget: it must be enough to produce conversions regularly — daily or near-daily, not a couple per week. Multiply your expected cost per conversion by a realistic event frequency and you get the minimum daily spend at which the system can actually learn. A business whose tolerable cost per lead is high but whose budget only affords one conversion a week isn't running a smaller version of a good campaign — it's running a campaign that never leaves the guessing phase. This floor depends entirely on clean conversion events reaching Meta, which is why tracking setup comes before budget setting, not after.

If your honest floor is more than you can sustain, don't split the difference by running half-budgets — see "when not to advertise" below.

Why "spend 10% of revenue" rules mislead

Percent-of-revenue rules survive because they're easy, not because they're right:

• They ignore margin. Two businesses at the same revenue can have opposite capacity to fund ads. The percentage flatters one and starves the other.
• They point backwards. Budgeting from current revenue means small businesses — the ones that most need new customers — get the smallest budgets, often below the learning floor.
• They have no success condition. A percentage tells you what to spend but not what result justifies it, so campaigns run on vibes and get cut the first slow week.
• They ignore capacity. A booked-solid business "should" spend the same percentage as a hungry one, which is obviously wrong.

Percent-of-revenue thinking has one legitimate use: as a sanity ceiling once the unit math already works. It should never be the starting point.

Stage the budget: learn, prove, scale

With the math done, don't commit the "final" budget on day one. Stage it, and attach a question to each stage:

1. Learn. Run at your learning-floor budget with a handful of genuinely different creative concepts. The question is not "is this profitable yet" — it's "which message and audience produce conversions at all, and at roughly what cost?" Judge nothing in the first days; buy a full testing cycle.
2. Prove. Hold spend steady on what survived, and confirm cost per real outcome — booked jobs, orders, consults — sits inside your tolerable acquisition cost, using your CRM or sales numbers rather than the platform's scoreboard. This is the stage most advertisers skip, and it's the one that makes scaling safe.
3. Scale. Raise budgets in steps, not leaps, and watch whether cost per outcome holds as spend grows. It usually rises somewhat — audiences and creative both fatigue — so scale until the marginal customer approaches your tolerable cost, then stop. The goal is maximum profitable volume, not maximum spend.

For local businesses there's an extra constraint at the scale stage — a finite radius saturates — which we cover in our guide to Meta ads for local businesses.

When the right budget is zero

Some situations shouldn't be advertised out of. Ours is a paid-media agency saying this:

• The margin isn't there. If honest math leaves almost nothing between customer value and delivery cost, ads amplify a pricing problem. Fix the offer first.
• The conversion path is broken. Slow site, nine-field form, phone that rings out — traffic into a leaky funnel is the most expensive way to learn your funnel leaks. A purpose-built landing page is usually cheaper than the spend it would waste.
• You can't sustain the learning floor. A budget below the floor, or one you'll pause in a panic after two slow weeks, buys confusion rather than data. Wait until you can fund a full learn-and-prove cycle.
• You're at capacity. If you can't serve more customers, spend the money on retention or on raising prices instead.
• Nothing has ever sold without ads. If an offer has no organic or word-of-mouth proof at all, ads become an expensive focus group. Small tests can be that focus group deliberately — just budget them as research, not as growth.

The honest answer

How much should a business spend on Meta ads? Enough to clear the learning floor, capped by what your margin math says a customer is worth, staged so each increase answers a question the previous stage raised, and zero until the funnel and the finances can carry it. That answer is less quotable than a dollar figure — and it's the one that keeps the money working for you instead of for Meta. If you'd rather not run the math alone, that calculation is the first thing we build in our Meta ads management engagements.