There is no single right number, but there is a right method: work backwards from what a new customer is worth to you, decide what you can afford to pay to win one, and then check that the monthly budget buys enough results for Meta's delivery system to learn. If the budget cannot buy that, a smaller test will mostly measure noise, and you may decide Facebook ads "don't work" on too little data.
Below is the calculation we use with small and mid-size businesses, a worked illustration, and the mistakes that waste the first month.
Start from what a customer is worth, not from a round number
Most owners pick a number that feels safe, such as $10 a day, and judge the channel by what happens. The problem is that the number has nothing to do with the economics of the business. Begin with three figures you already know or can estimate:
- Gross profit per customer. What a typical first order or job leaves after the cost of delivering it. If customers come back, use what you earn from them over a realistic period, such as 12 months, but be conservative.
- Close rate. Out of every ten enquiries or leads, how many become paying customers today? Use your own records, not an industry average.
- The most you are willing to pay to acquire a customer. This is a business decision, not a marketing one. Many owners choose a share of first-order profit so the first sale pays for itself; others accept a loss on the first order because repeat business is reliable.
From those you get your target cost per lead: maximum cost per customer multiplied by your close rate. If you can pay up to $300 to win a customer and you close three in ten leads, you can afford about $90 per lead.
Check the budget is big enough for Meta to learn
Meta's delivery system optimizes toward the event you choose, such as a lead or a purchase. Meta's own guidance is that an ad set needs about 50 of those events in a week after launch or after a significant edit to leave the learning phase. Below that, results tend to be less stable and harder to read.
That gives a second, independent number:
Weekly budget needed ≈ 50 × the cost per result you expect.
You will not hit 50 results a week on every account, and you do not need to. The point of the check is to tell you which goal is realistic:
- If 50 results a week at your expected cost fits your budget, optimize for the event you actually want, for example a qualified lead.
- If it does not, either consolidate into fewer ad sets so each one gets more data, or optimize for an event that happens more often and judge the campaign on the deeper event in your own records.
What not to do is spread a small budget over many ad sets, audiences and creatives at once. Each piece receives too little data to learn anything.
A worked illustration
The numbers below are made up to show the arithmetic. They are not a benchmark and not a promise of results.
A local renovation company:
- Gross profit on a typical job: $2,000
- Willing to pay up to: $400 to win a job
- Close rate from qualified enquiries: 1 in 5
- Target cost per qualified enquiry: $400 × 0.2 = $80
If the campaign optimizes for form submissions, and roughly half of submissions turn out to be qualified, the company can afford about $40 per form submission.
Learning check: 50 form submissions a week × $40 = $2,000 a week, or roughly $8,600 a month. That is more than this company wants to spend. So instead of five ad sets for five services, it runs one campaign with one or two ad sets for its most profitable service, accepts a slower learning phase, and judges results on cost per qualified enquiry from its own lead list rather than on what Ads Manager calls a lead.
At around $1,500 a month, that company could expect, at the target cost, roughly 35–40 form submissions and 18–20 qualified enquiries a month, which is enough to see within a month or two whether the channel can work at its numbers.
How long a fair test takes
Give a new campaign at least two to four weeks before deciding, and avoid large edits in the first week, because each significant change can restart learning. Decide in advance what you will measure: cost per qualified lead or cost per sale, not clicks or reach. If after a fair test the cost per customer is well above what you can pay, the fix is usually one of four things, roughly in this order:
- The offer. Is it clear who it is for, what it costs or roughly costs, and why act now?
- The landing page or form. Does it filter out people who cannot buy, and is it fast on a phone? Our notes on landing page anatomy cover what moves conversion rate.
- The signal. Is Meta optimizing for an event that correlates with customers, or for any form fill? See leads vs qualified leads for this problem in detail.
- The creative. Enough distinct concepts tested in a structured way; see creative testing systems.
Only after those should you conclude that the budget itself is the problem.
Common budget mistakes
- Daily budget too small for the goal. $5 a day optimizing for purchases of a $40 product rarely produces enough purchases for the system to learn.
- Too many ad sets. Ten ad sets at $10 a day each learn far more slowly than one or two at $50 a day.
- Boosting posts instead of running campaigns. Boosted posts are simple, but they offer fewer options for the objective and optimization event. For lead generation or sales, a campaign built in Ads Manager gives you more control over what Meta optimizes for.
- Judging on cost per click. Cheap clicks from people who never buy are expensive.
- Turning ads on and off. Stopping and restarting, or changing budgets sharply every few days, keeps the campaign in an unstable state.
- No tracking of what happens after the lead. If you cannot tell which leads became customers, you cannot tell which ads are worth the money.
When it makes sense to bring in an agency
Doing it yourself is reasonable when the budget is small, the offer is simple and you have time to learn the tools. An agency becomes worth considering when the ad budget is large enough that mistakes are expensive, when tracking from lead to sale is missing, or when nobody in the business has time to test creative every week. We wrote a longer comparison in agency vs in-house media buying.
How we work
We start from your numbers: what a customer is worth, your close rate and how leads are recorded today. From there we size the test, choose one optimization event and set up tracking so results are judged on qualified leads or sales, as described on our Meta Ads service page. We work with businesses that have a website or an active business page and an ad budget from $1,000 a month, paid to Meta separately. We do not promise a cost per lead before seeing your account and funnel.
If you want a second opinion on your budget, apply here with your monthly ad spend, what a customer is worth to you and how leads are recorded. We reply in writing with how we would size the first test.
Frequently Asked Questions
What is the minimum budget for Facebook ads?
Meta lets you run ads on a few dollars a day, so the technical minimum is low. The practical minimum is the budget that buys enough results to learn from: roughly 50 optimization events a week at your expected cost per result is Meta's guidance for leaving the learning phase.
Is $10 a day enough for Facebook ads?
It can be enough to test whether people engage with an offer, or to retarget a small audience. It is rarely enough to optimize for leads or purchases when each one costs more than a few dollars, because the campaign will not collect enough results to stabilize.
Should I spend my budget on Facebook or Instagram?
Usually you do not need to choose. One Meta campaign can deliver across Facebook, Instagram and other placements, and the system shifts spend to where results are cheaper. Split them only if your creative only works in one format.
How long before I know if Facebook ads work for my business?
Plan on at least two to four weeks of stable spend with tracking from lead to sale in place. Judge on cost per qualified lead or cost per customer, not on clicks or reach.
Does the ad budget include the agency fee?
No. The ad budget is what you pay Meta for delivery. Any agency or freelancer fee is separate and should be agreed in writing before work starts.