How much should you spend on Meta ads?

Start with what a customer is worth

Before any formula, work out three numbers: what a new customer is worth to you after costs (your margin), what share of your leads usually become customers, and therefore the most you can afford to pay for a lead or a sale.

A hypothetical example in round numbers: if a new customer brings you $600 of margin and about 1 in 10 leads buys, a lead is worth up to $60 to you before you make a loss. If ads are likely to bring leads at around $20, there's room to work. If they'd cost $80, the plan needs a different offer, price or channel before it needs a bigger budget. All amounts in this guide are US dollars.

Why "how much" depends on what you optimise for

Meta will happily spend $5 a day for you. The real question is whether that budget lets its delivery system learn. Every ad set starts in a learning phase, where Meta is still working out who to show your ads to. Results are unstable until it has seen enough of the action you asked it to optimise for.

Meta's help pages call an ad set learning limited when it's unlikely to get around 50 optimisation events in the week after its last significant edit. Treat 50 as a planning number for your setup, not a law and not a spending target. Meta changes these systems often, so check the current help page (linked below) before you plan.

The formula

Daily budget per ad set ≈ (50 × expected cost per result) ÷ 7

If you don't know your cost per result yet, use a cautious estimate from your industry or from anything you've run before, then correct it after the first couple of weeks. Some worked examples, with round hypothetical numbers:

• Leads at $8 each: (50 × 8) ÷ 7 ≈ $57 a day
• Leads at $20 each: (50 × 20) ÷ 7 ≈ $143 a day
• Purchases at $60 each: (50 × 60) ÷ 7 ≈ $429 a day

This is per ad set. Two ad sets need twice the money to learn at the same speed. That's the most common budget mistake I see in small accounts.

Why splitting a small budget backfires

It feels sensible to test many audiences at once, each with a little money. In practice each ad set gets a handful of results a week, none of them ever finishes learning, and the account looks like it doesn't work. Spreading a budget thinly also tends to shrink how many people you reach while the same people see your ads more often. That's a sign of the structure, not of bad targeting.

One ad set with a few good ads, left alone for several weeks, very often beats a dozen small ones.

What to do when the formula says more than you have

Many businesses fall below the number, and that's fine. You can still advertise well; you just set it up differently:

1. Use one campaign and one ad set. Put the whole budget where Meta can learn from all of it.
2. Consider a more frequent action. Meta itself suggests this for learning-limited ad sets, for example add to basket instead of purchases. The trade-off is real: Meta will find people who take that action, and they aren't always the people who buy. If you switch, keep watching purchases or qualified leads, and only keep the change if those hold up.
3. Keep targeting broad and let your creative do the targeting. Small budgets don't leave room for narrow audiences.
4. Agree a review plan before you start: what you'll check, and when. Leave the setup alone for roughly two to three weeks unless something is clearly wrong, such as broken tracking, rejected ads or spend running ahead of what you can afford. Fix those straight away.
5. When something works, raise the budget gradually. Around 20% every few days is a widely used safety margin. It isn't an official Meta rule, but big jumps often knock an ad set back into learning.

How long before you can judge results

It's driven by results, not by calendar days. Divide your weekly budget by your cost per result. If that comes to well over 50, one week can tell you something. If it's 10 or 15, which is normal for a small account, you need two or three weeks before the numbers mean anything. Judging after five days at $30 a day is reading noise.

A sensible starting budget

I don't believe in one magic number. My own approach is to plan for at least three to four weeks of spend before deciding anything, to only spend money you can treat as the cost of learning, and to set the point at which you'd stop or change course before you start. If you'd like a budget worked out for your own numbers and market, that's part of building a marketing strategy.