A small business advertising budget is not a number you look up, it is a floor you calculate. The starting point is not a percentage of your revenue but your cost per result and the volume a platform needs to optimise. Meta advises a daily budget of at least 5 times your cost per result, and Google recommends at least 3 times your CPA for a Performance Max campaign. So the useful question is not “what should I spend”, it is “what does one result cost me, and how many do I need before the numbers mean anything”.

That answer is unusual, because almost every page on this topic replies with a percentage or a round monthly figure. A quick sweep of US advice turns up recommendations from 1 percent to more than 11 percent of revenue, and from $100 to several thousand dollars a month, for businesses described as comparable. A figure that comes from a method does not swing that widely.

How much should a small business spend on advertising?

The most repeated answer is a share of revenue, and it is the least useful one. You will read that a small business should put 7 to 8 percent of revenue into marketing, a rule very often attributed to the US Small Business Administration.

That attribution does not survive a check. The SBA’s own page on marketing budgets states plainly that “there’s no hard and fast answer” and, far from recommending 7 to 8 percent, cites an average advertising spend of 1.08 percent of revenue. The larger marketing figures that circulate cover total marketing, including salaries, tools, agencies and events, of which paid advertising is only a fraction. We take apart the percentage myth in detail in our guide to marketing budget as a percentage of revenue.

The deeper problem is that no percentage of revenue can tell you whether a campaign will be profitable for you. Two businesses with identical revenue can face wildly different lead costs. What decides your floor is the cost of one result and the volume a platform needs to learn, not a slice of your top line.

How much per month? The learning-phase floor

Your monthly floor is set by a single mechanic: below a certain volume, the platform’s algorithm never learns. This is the most misunderstood point in online advertising, and the one that explains most failed first campaigns.

When you launch a campaign, the platform runs a learning phase, testing audiences, placements and timing to find what converts. Meta states that ad sets exit this phase “after about 50 [optimisation events] in the week following the last significant edit”. Note the wording: about, typically. It is an order of magnitude, not a switch that flips at exactly 50.

That threshold turns into a budget as soon as you multiply it by your cost per result. If a lead costs you $27.66, the 2025 US average measured by WordStream across 726 US campaigns, then 50 leads a week costs roughly $1,383, or about $6,000 a month for a single ad set. This is our calculation applied to a published benchmark, not a figure Meta publishes.

The monthly learning-phase floor scales with your cost per resultTable with three rows illustrating how the monthly advertising floor rises with the cost of one result, on the assumption that a Meta ad set needs about 50 results per week, or roughly 217 per month, to exit the learning phase. First row, a low cost per result of about 3.16 dollars, seen in cheap niches such as restaurants: about 217 results per month, giving a monthly floor of about 700 dollars. Second row, the 2025 US average lead cost of 27.66 dollars measured by WordStream: the same 217 results per month give a monthly floor of about 6,000 dollars for a single ad set. Third row, a high cost per result of about 70 dollars, seen in legal and finance: the same volume gives a monthly floor of about 15,200 dollars. The conclusion is that a fixed monthly budget such as 500 dollars only clears the learning phase when the cost per result is very low, and that the floor is a calculation from unit cost, not a round figure.The monthly floor is a calculation, not a round figureAssuming about 50 results per week to exit Meta’s learning phase, roughly 217 per month for one ad setCost per resultResults per monthMonthly floor$3.16 low, cheap nichesabout 217about $700$27.66 US average leadabout 217about $6,000$70 high, legal and financeabout 217about $15,200A fixed $500 a month only clears learning when a result costs roughly $3 or less.At the US average lead cost, the same $500 buys about 18 leads a month, well below learning volume.Figures are our calculation. Meta uses the words “about” and “typically”; the 50 results are an order of magnitude.
The floor to exit Meta’s learning phase scales directly with your cost per result. These figures are our calculation, applying the roughly 50 results per week Meta cites to three published lead costs. Source : Our calculation, from WordStream 2025 and Meta (2025)

Two cautions before you apply this. Meta’s 5 times rule is tied to one specific bid strategy, the cost per result goal, so it is not universal. And Google frames its 3 times figure as a recommendation, “an average daily budget of at least three times your CPA”, not a technical limit. Neither platform publishes a minimum in dollars: the number only appears when you create the campaign.

Is $500 a month enough for Facebook ads?

It depends entirely on what one result costs you, and for most lead generation the honest answer is no. At the 2025 US average of $27.66 per lead, $500 a month, roughly $16 a day, buys about 18 leads a month. That is a long way below the 50 results a week an ad set needs to leave the learning phase, so the spend funds an exploration that never converges.

The range is what settles the question. WordStream’s 2025 data puts the leads objective between $3.16 and $76.71 depending on industry, so $500 a month is generous for a restaurant at the low end and nowhere near enough for a dental or legal practice at the high end. The number that decides whether $500 works is your cost per result, not the $500 itself.

There is a separate figure that confuses this debate: the platform’s technical minimum, which is very low, on the order of a few dollars a day. That minimum lets you switch a campaign on. It does not let it learn. Facebook’s lead cost also keeps climbing, up 20.94 percent year over year in WordStream’s 2025 report, which pushes the same $500 to buy fewer leads each year at equal performance. We break the unit economics down in our guide to how much Facebook ads cost.

How to calculate your own advertising floor

Start from your own commercial figures, not an industry table. You already know them, even if you have never written them down this way.

The method is four questions:

  1. What is a customer worth to you? Take the margin on an average deal, not the revenue.
  2. How many enquiries or meetings does it take to close one? Your sales team knows this.
  3. What can you pay for one inbound enquiry? That is your margin per deal divided by the enquiries needed, then divided by two or three to keep the deal profitable.
  4. Apply the platform multiple. Multiply that figure by 5 for Meta or 3 for Google, then by the learning volume, and you have your monthly floor.

This beats any sector benchmark because it rests on numbers you can verify inside your own business. To set it up cleanly, including the costs most people forget, follow our method for calculating customer acquisition cost.

B2B versus B2C: why the number is not the same

The right budget is not the same for a B2B and a B2C business, and the gap runs in a way that surprises people. On figures cited by the SBA, US marketing budgets skew higher for consumer businesses: B2C services near 11.8 percent of revenue against 6.9 percent for B2B services, with B2C products at 9.6 percent and B2B products at 6.3 percent. Read those as total marketing budgets, not paid advertising alone.

You might conclude that B2B should therefore spend less on ads. The opposite is usually true at the campaign level. B2B tends to have a higher cost per lead and a longer sales cycle, so it takes more spend, not less, to reach the 50 results a week that make a campaign readable. A B2B advertiser with a $120 cost per lead needs a far larger concentrated budget to learn than a B2C advertiser at $15. That is why a modest B2B budget spread across search, social and display almost always fails on all three. We set out how to structure this in our guide to B2B paid acquisition.

What if your budget is below the floor?

Concentrate it, or wait. Those are the only two honest options, and spreading an insufficient budget is not one of them.

Concentrating means shrinking the scope until the budget is enough again on that smaller scope: one offer instead of your whole catalogue, one region instead of the whole country, one channel instead of two. A modest budget on a single metro area can reach learning volume where the same budget nationwide never will.

Waiting is sometimes the right call, and it is rarely written down. If even a concentrated budget sits well under the floor, paid advertising is not your next investment. Organic search, referrals, local partnerships or a rebuilt website will return more per dollar until your cost per lead is known and your funnel is in order. This is not a rare position. The average small business advertising budget looks large, around $78,000 a year in Intuit’s 2025 survey of 1,006 owners, precisely because a minority of heavy spenders pulls the average up while most owners spend far less. If you have not launched a campaign yet, you are not behind your market, you are in the majority.

A final technical point that causes needless worry: both platforms can spend above your daily budget on a given day. Meta can run up to 75 percent over on one day within a weekly cap, and Google can spend up to twice the daily average on a day. Your monthly total is still respected, even when a daily report surprises you.

In short

  • Calculate, do not estimate. Your floor is your cost per result times the volume a platform needs to learn, roughly 5 times cost per result a day on Meta or 3 times CPA on Google, not a percentage of revenue.
  • $500 a month only clears learning if a result costs about $3. At the US average lead cost of $27.66, that budget buys about 18 leads a month, well short of the 50 a week an ad set needs.
  • Concentrate before you widen. One offer, one region, one channel, until you pass learning volume. That threshold, not the headline budget, separates useful spend from wasted spend.

Reading your real cost per result and building a floor around it is the core of our B2B paid acquisition work. If you would rather test these numbers against your own margin and conversion rate before committing anything, book a diagnostic. We look at your margin per deal, your close rate and the channel that fits your market, and you leave with a calculated floor, even when the honest conclusion is to wait.