Enough to get about 50 conversions a week on a single campaign, and nothing spread thinner than that. Meta's delivery system needs roughly that volume before it optimizes properly, so your floor is 50 multiplied by what one conversion costs you. Percentage-of-revenue rules are a sanity check, not an answer.
That is a less satisfying answer than "spend 7% of revenue", and it is the one that survives contact with an actual ad account. The percentage rules were built by surveying large companies. You are probably not one.
What do the budget benchmarks actually say?
They say what large marketing departments do, which is useful context and a bad instruction.
The CMO Survey puts overall marketing budgets at 9.0% of company revenues in 2026. Gartner's CMO Spend Survey, measuring a different population with a different method, puts it at 7.8%, up from 7.7% in 2025. Of that marketing budget, social media spending accounted for 11.3% in early 2025, down from 12.1% in Fall 2024.
Run the arithmetic on a small business and watch it fall apart. A shop doing $600,000 a year, at Gartner's 7.8%, has a $46,800 marketing budget. At 11.3% of that, social ads get $5,288 a year, or about $440 a month.
That $440 sounds reasonable until you ask what it buys. If a conversion costs you $25, that is 17 conversions a month across every campaign you run. Meta wants 50 a week in a single ad set. You are not underfunded by a little. You are below the threshold where the system can work at all.
The benchmark is not wrong. It is measuring companies where 11.3% of the marketing budget is a real number.
What is the actual floor?
Meta publishes it, which not enough people notice.
An ad set enters what Meta calls "learning limited" when it is unlikely to receive around 50 optimization events in the week following your last significant edit. Once it does receive 50, it moves out of that state and costs become more stable. That is Meta's own documentation in its Business Help Center, not an agency rule of thumb.
So your minimum viable weekly budget is:
50 × your cost per optimization event
If a lead costs you $20, that is $1,000 a week to run one ad set properly. If you are optimizing for a cheaper event, such as a landing page view or an add to cart, the number drops accordingly, and so does the quality of what the system learns.
This is the single most useful calculation in paid social and almost nobody does it before setting a budget.
What if I cannot afford the floor?
Then you change the event you optimize for, or you do not run ads yet. Those are the two honest options.
Optimizing for a cheaper, earlier action gets you volume the system can learn from. A campaign optimizing for landing page views at $1.50 hits 50 events on $75 a week. The trade is that you are teaching Meta to find people who click, not people who buy, and those are different people.
That trade is worth making when you are starting and have no conversion data at all. It stops being worth making the moment you can afford to optimize for the thing you actually want.
What does not work is running a purchase-optimized campaign on $200 a month and concluding that Facebook ads do not work. You never gave the system enough signal to find out.
The three tiers
| Tier | Monthly spend | Structure | What it is for |
|---|---|---|---|
| Starter | Floor × 4.3 for one ad set | One campaign, one ad set, two or three creatives | Finding out whether paid works for you at all |
| Growth | Two to three times the starter tier | One prospecting campaign, one retargeting campaign | Scaling something already proven, with a warm audience catching what prospecting misses |
| Aggressive | Whatever the unit economics justify | Multiple prospecting angles, retargeting, ongoing creative testing | Volume, when you know your numbers and the constraint is production, not budget |
Notice that the tiers are defined by structure rather than by a dollar figure. A dental practice and an ecommerce brand can be in the same tier at very different spends, because the floor is set by their cost per conversion, not by their revenue.
The jump most businesses get wrong is starter to growth. They double the budget and also double the number of campaigns, which leaves every campaign exactly as starved as it was before, just twice as expensive in aggregate.
How do you know if it is working?
By whether it pays, which is a different question from whether the metrics look good.
Work out what a customer is worth to you over the life of the relationship, not on the first purchase. If a client is worth $2,000 over three years and you are acquiring them for $180, the campaign is working even if the cost per lead looks alarming next to a benchmark. If a customer is worth $40 once and you are paying $35 to get them, it is not working no matter how good the click-through rate is.
Benchmarks cannot tell you this because benchmarks do not know your margins. I laid out why published cost figures vary so wildly, and how far to trust them, in the Meta ads comparison.
What I actually see in small accounts
The pattern is remarkably consistent, and it is nearly always structural rather than financial.
The account has six campaigns running on a budget that would properly fund one. Each has been on for months, each sits in learning limited, and none has ever produced stable costs. The owner concludes the targeting is wrong and adds a seventh campaign.
The clearest case I can point to is an RV dealer whose account we took over. We cut the running creative set down to two images and one video, and refreshed that set every few weeks instead of stacking new assets on top of it. The budget did not change. What changed was how many places it had to be at once.
That is the same arithmetic as the campaign count, one level down. Every additional asset divides the same spend into smaller portions, and a portion sitting below the learning threshold never reaches stable costs no matter how long you leave it running.
The fix is arithmetic, not strategy. Add up the total spend, divide by the cost per conversion, and see how many events the account can actually generate in a week. If the answer is under 50, you have one campaign's worth of budget regardless of how many campaigns are currently switched on.
Getting this right is most of what the paid team at Inertia Digital Marketing does in the first month of any engagement, and it usually involves turning things off rather than on.
What to do on Monday
Work out your cost per conversion from the last 90 days. Multiply it by 50. That is your weekly budget for one properly funded ad set.
Compare it to what you are currently spending. If your spend is above that number, you can afford more than one campaign. If it is below, you can afford exactly one, and every additional campaign you are running is taking budget from it.
Then decide whether the number you need is one you can justify against what a customer is worth. If it is not, the answer is not a smaller ad budget. It is no ad budget, and a different channel, until the economics change.
For the wider question of how paid and organic split across your whole strategy, start with the paid versus organic pillar.
FAQ
How much should a small business spend on social ads? Start from your cost per conversion rather than a percentage of revenue. Meta's delivery system needs roughly 50 optimization events per ad set per week to leave the learning phase, so multiply your cost per conversion by 50 to get a weekly floor for one properly funded campaign.
Is there a minimum budget for Facebook ads to work? Effectively yes, though Meta does not publish a dollar figure. The constraint is conversion volume, not spend. If your budget cannot generate about 50 optimization events a week in one ad set, costs stay unstable and the system never optimizes properly.
What percentage of revenue should go to marketing? The CMO Survey puts overall marketing budgets at 9.0% of revenue in 2026 and Gartner puts it at 7.8%. Those figures describe larger companies and make a poor instruction for a small business, where the same percentage often produces a budget below the point at which paid social can function.
Should I spend more on ads or on content? If nobody knows you exist, ads. If people find you and are not convinced, content and your website. Extra ad spend sends more people to the same experience, which multiplies whatever is already happening, good or bad.
Why did my ads stop working when I increased the budget? Usually because the increase triggered a significant edit and pushed the ad set back into the learning phase, or because the extra budget was used to launch more campaigns rather than fund the existing one. Raise budgets gradually and resist adding campaigns at the same time.