How Much to Spend on Facebook Ads for a Small Business

You are ready to try Facebook ads, and every source gives you a different number. One person says five dollars a day is plenty, an agency proposal says a thousand a month, and a video says anything under fifty a day is wasted. None of them explain where the number came from, so you have nothing to say when a partner or a bookkeeper asks why you picked it. Underneath the question of how much to spend on Facebook ads is a simpler fear: paying for something that brings back nothing. The right figure for your business can come from your own numbers, and you probably have most of them already. Once you have them, you can set a starting budget, a test length and a hard limit on what you are willing to lose. You will know before you spend the first dollar what result would make you stop, change course or keep going.

Your Facebook ad budget starts with what one new customer is worth

There is no single dollar amount that is right for every small business, and Meta does not publish one. Its help page about budgets explains how daily and lifetime budgets work, and Ads Manager applies a minimum that can change with your currency, your billing event and what you ask the campaign to optimize for. That minimum is what the system will accept. It is not the amount that tells you whether ads bring you customers.

The number that matters is how much you can afford to pay for a lead and still make money on the customers who come from it. That depends on your margin, your prices and how often your leads turn into paying work, and nobody outside your business knows those. A plumber, a dentist and a dog groomer can run the same ad for the same week and need completely different budgets to learn anything.

So, the question to answer first is not "how much per day?" It is "how much can one lead cost me before it stops being worth it?" Once you have that ceiling, the daily and monthly figures follow from simple division.

The work-back method turns your own numbers into a cost-per-lead ceiling

The method has four steps, and each one uses a figure you can pull from your own records. You start with profit from one customer, decide how much of it you will spend to win that customer, and then account for the leads that never buy. The result is a maximum cost per lead, and from there a test budget and a loss cap. None of it requires a benchmark or a guess about the auction.

Step one: find the contribution margin from one new customer

Take the average revenue a new customer actually pays you, then subtract direct costs: materials, labor on the job, refunds and any sales commission. What is left is contribution margin. Use this, not the ticket price, because the ticket price includes money that was never yours to spend on ads. If a typical customer buys more than once in the period you are willing to fund, you can count those purchases too, as long as you have the records to back it.

Step two: decide what that margin can pay to win a customer

Choose the share of that margin you are willing to hand to advertising. Set aside a reserve for your own sales time and any creative or management fees first. The dollar figure you land on is your maximum acquisition cost per new customer. Spend more than this to win one customer and each sale loses money, even if the ads look busy.

Step three: multiply by your real close rate

Look at your last 20 to 50 genuine leads and count how many became paying customers. That share is your close rate. Multiply your maximum acquisition cost by it, and you have your maximum cost per lead. If one lead in five buys, you can afford to pay one fifth of your acquisition cost for each lead.

Step four: turn the ceiling into a test budget and a loss cap

Decide how many qualified leads you need to see whether your close rate holds, then multiply by your cost-per-lead ceiling. That is your media allowance for a four-week test. Divide it by 28 for a daily figure. Then set a loss cap: the most you can lose on the test without touching payroll or the ability to serve customers. If the allowance is bigger than the cap, the cap wins and the daily budget comes down.

These are illustrations of the arithmetic, not recommended budgets. Your margin and close rate will be different, and the cost-per-lead ceiling in Example A comes before any ad-management fees. If you have no reliable margin or close-rate records, run a deliberately smaller first test and track calls, appointments and sales, rather than assuming a published cost per lead is one you can afford. The SBA's guide to calculating costs is a useful way to separate one-time costs from the monthly ones that belong in your margin.

Meta's learning phase explains why $5 a day often cannot answer the question

When you launch a new ad set, or make a big change to one, Meta puts it into a learning phase. Its help page about the learning phase says performance is less stable during this period, while the system works out who to show your ads to. Meta's long-standing planning guideline is about 50 optimization events per ad set within seven days to leave learning. An optimization event is whatever you told the campaign to go after, such as a lead form submission.

That guideline gives you a way to estimate how much Facebook ads cost per day for a given goal. Multiply the cost you expect for each event by 50, then divide by 7. At a $25 cost per lead, that is about $179 a day for one ad set. At the $40 ceiling from Example B, it is about $286 a day.

Compare that with a $5 a day test. Over four weeks it spends $140. At $25 a lead, that buys five or six leads if everything goes well. That can still tell you something useful, such as whether anyone stops to look at your ad at all. It normally cannot tell you what a qualified lead really costs for an expensive local service, and it will not leave the learning phase.

Two limits keep this in proportion. The 50-event figure is a guideline for stable performance, not proof that a campaign below it cannot produce a profitable sale. And an "In learning" label means results may swing more than usual. It is not an instruction to spend past your loss cap. If your budget is small, the practical move is to put all of it into one campaign and one ad set, rather than splitting it across several audiences, placements or goals, where each would collect even fewer events.

Published benchmarks are comparators, not a forecast

Once your test is running, it helps to know whether your cost per lead is roughly normal or wildly off. Published benchmarks can tell you that, as long as you read them for what they are. The most recent public US source in the research for this post is the LocaliQ and WordStream Facebook advertising benchmarks for 2025. LocaliQ and WordStream sell advertising services, so this is data an advertising company published about the campaigns it saw. It covers 726 US campaigns from April 2024 through June 2025.

Read these figures with their limits beside them. They are campaign medians, not a quote for your town or a prediction of what you will pay. The report does not break out cost per thousand impressions (CPM) for these local categories, and it does not separate Instagram from Facebook. Your audience size, offer, season, placements, what you count as a lead and how quickly you follow up can all move your cost a long way from any median. WordStream's 2024 benchmark guide is useful as a prior-year comparison, with the same caution.

The use for a benchmark is spotting an extreme. If your home-improvement leads cost several times the category median, something in the setup deserves a look. If they cost less, that still tells you nothing until you know how many of those leads booked. Your own ceiling decides whether a cost per lead is good. The benchmark only tells you whether it is unusual.

A first test runs four weeks against a loss cap you set in advance

Four weeks is a practical window for a first test, long enough to get past the first week of unstable results and short enough to limit what you can lose. It is a decision window, not a Meta rule. Your Facebook ads cost per month for the test is simply your four-week allowance, or your loss cap if that is lower. Write both numbers down before launch, along with what counts as a qualified lead and who answers the leads.

Keeping the test clean is what makes the result usable. Meta's page on significant edits and the learning phase explains that big changes send an ad set back into learning. If you change the audience on Tuesday, the image on Thursday and the offer on Saturday, you lose the ability to say which change did what, and you restart the unstable period each time. Change one major variable at a time, then give it room.

Stop, change and scale rules decide what happens after the numbers arrive

A test is only useful if you already know what each result will make you do. Without rules set in advance, the choice to raise, cut or cancel the budget tends to follow your mood that week rather than your numbers. The three rules below come straight from the ceiling and cap you have already set. They apply to a lead campaign for a local service, where the result you care about is booked work.

Stop

Stop early if the setup is broken: ads are rejected, the lead form or phone path does not work, your tracking counts the same lead twice or records spam, or there is a compliance problem with what the ad claims. If Meta restricts the account itself, the steps in where to look first when a Facebook ad account is restricted come before any budget decision. Stop at the end, too, if spend reaches your loss cap with no qualified leads and no clear repair to try.

Change

Change one thing when the ads are running but the numbers miss. Candidates are the offer, the creative, the form or landing page, the audience or the optimization event. Pick the one the evidence points at. Leads that never answer the phone point at the form or the follow-up. Plenty of clicks and no leads point at the page or the offer.

Continue and scale

Continue when leads meet your qualified-lead definition and your cost per lead sits at or below your ceiling, and you have followed up long enough to see appointments or sales. Scale only when you can answer the extra calls quickly and those early leads are turning into paying customers. Raise the budget in measured steps, watching cost per qualified lead and cost per sale. A falling cost per click on its own is not a reason to scale, because cheap clicks that never book are still a cost.

A fifteen-minute check comes before any budget

You can do this at the kitchen table before you open Ads Manager, with a notepad and your recent sales records. It works the same for Facebook ads for a local business as for one that sells further afield, because it uses only your numbers.

• Minutes 1 to 5: Take the average revenue from a new customer and subtract direct costs, refunds and commissions. That is your contribution margin.
• Minutes 6 to 10: Count your last 20 to 50 genuine leads and how many became paying customers. Customers divided by leads is your close rate.
• Minutes 11 to 15: Multiply your margin by the share you will spend on acquisition, then by your close rate. That is your maximum cost per lead. Write down a four-week loss cap you could absorb without strain.

If you cannot fill in the close rate, that is the next job, and it matters more than picking a budget. Sit down with whoever answers your inquiries and go through the recent ones. Then choose one outcome to measure, such as a qualified call or a booked estimate, and make sure you can record where each one came from. Meta's Pixel and Conversions API connect website actions to your ads, and a simple sheet that logs every lead and what happened to it does the rest.

When a test disappoints, check the causes outside the auction before blaming it. No national ranking of failure causes exists, so treat this as a diagnostic order rather than survey data: whether qualified leads and sales are defined and tracked, how fast leads get a reply, whether the offer and form give people a clear reason to reach out, whether a small budget is split too thin, whether the creative matches the offer, whether the campaign goal is right, targeting and policy limits, and whether you can serve the demand at all.

Some worries matter less than they seem

A few things take up more of your attention than they deserve. You do not need separate budgets for Facebook and Instagram before your own results show a reason. You do not need to narrow your audience to a sliver before there is evidence it improves lead quality; in an auction, extra limits can shrink the room Meta has to find people. Likes, follows and cheap traffic are not the measure of a lead campaign.

What does deserve attention is what your ads say. The FTC's advertising FAQs for small businesses say ads must be truthful and not misleading, and its policy statement on advertising substantiation, dated 1984, says you need a reasonable basis for objective claims before the ad runs. That covers prices, savings, results and testimonials. It does not set a budget. No regulator or standards body publishes a recommended Facebook ad spend. If you advertise housing, employment or credit, Meta's Special Ad Categories rules apply, and sector-specific legal advice is worth getting before launch.

Frequently Asked Questions About how much to spend on facebook ads

How much should I spend on Facebook ads per month?

Set it from your own cost-per-lead ceiling. Multiply that ceiling by the number of qualified leads you need to judge a test, and cap it at a four-week amount you could lose without strain. Meta does not publish one right monthly figure.

How much should I spend on Facebook ads per day?

Divide your four-week test allowance by 28. To estimate what one ad set would need to leave Meta's learning phase, multiply your expected cost per lead by 50 and divide by 7.

Is $5 a day enough for Facebook ads?

It can buy ads and show whether people respond to them. For an expensive local service, four weeks at $5 a day may not bring in enough qualified leads to judge what a lead really costs.

What is a good cost per lead on Facebook?

One below your own profitable ceiling. Published medians, such as the LocaliQ/WordStream 2025 report's $27.66 across all industries for lead campaigns, are dated comparators from an advertising company, not a target.

How long should I test Facebook ads?

Allow about seven days after launch or a big edit before judging, then use a four-week window and a loss cap you set in advance to decide whether to stop, change one thing or continue.

Why are my Facebook leads poor quality?

Check the offer, whether the form asks enough to qualify people, who the ads reach, how leads are routed and how fast someone replies, before judging the cost per lead itself.

Final Thoughts

How much to spend on Facebook ads is a question your own records can answer. Start with the margin from one new customer, decide what share of it you will spend to win that customer, and multiply by your real close rate to get a maximum cost per lead. Meta's learning guideline shows why a very small daily budget may not produce enough leads to judge, and published benchmarks show whether your results are unusual. Neither one replaces your own ceiling.

With a ceiling, a four-week window, a loss cap and written stop, change and scale rules, the money you spend buys a decision instead of a guess. Whatever the test shows, you will know why, and you will have a number you can explain to anyone who asks.

If you would like a second pair of hands on the setup, Web Leveling can help you turn those numbers into a working campaign. Our Facebook ads management work starts from your cost-per-lead ceiling, keeps the test clean and reports qualified leads and booked work alongside spend, in an ad account that stays in your name. We work with small and medium businesses across the country and overseas. Send us your margin, close rate and the budget you have in mind, and we will help you set up a test you can defend.

Terms