How Much Daily Budget Do You Need to Start Advertising on Facebook in 2026?

Ask five people how much you need to spend on Facebook ads and you will get five numbers: a dollar a day, five dollars a day, twenty, fifty, “at least a thousand a month.” They cannot all be right, and the reason they contradict each other is not that some of them are lying. It is that they are quietly answering different questions.

This piece gives you the number for your campaign, and it does it by showing the arithmetic rather than the range. The short version is that Facebook has three different budget floors, only one of which is printed anywhere in the interface, and the one that decides whether your money works is a function you can calculate in about ten seconds.

The short answer

Meta will take about $1 a day, and suggests starting at around $5 a day over at least six days. But the budget that actually works is set by the learning phase, not by billing. An ad set needs roughly 50 optimisation events in a 7-day window before Meta stops guessing. So:

Your real daily floor = (target cost per result × 50) ÷ 7

Run that on the 2026 medians and you get two very different answers from the same formula:

If you optimise for	2026 median cost	Learning floor per day	Per 30 days
A link click (traffic)	$0.70	$5	~$150
A click on a lead campaign	$1.92	$14	~$411
A submitted lead form	$27.66	$198	~$5,927
A lead, dental practice	$76.71	$548	~$16,438
A lead, restaurant	$3.16	$23	~$677

Median costs are from WordStream’s 2025 Facebook Ads benchmarks, the most recent large public dataset — 554 traffic campaigns and 726 lead campaigns run in the US between 1 April 2024 and 30 June 2025, reported as medians.

That table is the whole article. The “$5 a day” advice is true — for traffic campaigns. The “you need thousands” advice is also true — for lead campaigns. Both fall out of one formula, and the variable that separates them is not your budget, your niche or your ad creative. It is which optimisation event you picked in step one of campaign setup.

The three floors, only one of which is visible

Every platform has a billing minimum. Almost nobody tells you that it is the least useful number in the account.

Floor 1 — the billing floor. What Meta will physically accept. Meta’s own budget guidance recommends starting “with at least a $5 budget” over “at least a 6-day duration” so the system can find the right audience; impression-optimised ad sets will run for around a dollar a day. This floor exists to let anyone start. It is not a recommendation about whether the money will do anything.

Floor 2 — the learning floor. What Meta’s delivery system needs to stop guessing. This is the one that matters, and Meta documents it plainly in the learning limited help article: an ad set becomes learning limited when it is unlikely to receive around 50 optimisation events in the week after your last significant edit. Meta is explicit that this is not a penalty — it is a signal that your budget is not being spent effectively, because the delivery system cannot optimise on the data it is getting. Their listed fixes are to raise the budget, adjust the bid strategy, or choose an optimisation event that occurs more frequently.

Floor 3 — the decision floor. What you need to make a confident call. Fifty conversions is enough for Meta’s model to find a pattern. It is thin for a human deciding whether to keep spending. If you are comparing two creatives and one produces 12 leads and the other produces 9, you have learned essentially nothing — that gap is comfortably inside random variation. In practice, a decision you would actually bet on needs somewhere in the region of 100 events per variant, which doubles the learning floor whenever you are testing rather than simply running.

Most budget advice online quotes Floor 1 and lets you assume it behaves like Floor 2. That single substitution is responsible for a very large share of the money wasted on Facebook by small advertisers.

Why every article you have read gives a different number

Because “cost per result” is not a property of Facebook. It is a property of the result you chose.

Run the formula backwards and the contradictory advice resolves neatly:

• Someone who says “$5 a day is fine” ran traffic or engagement campaigns. At $0.70 a click, $5 a day is 7 clicks a day, 50 a week. The maths works.
• Someone who says “$20–50 a day minimum” ran conversion campaigns for a cheap event — an add-to-cart, a landing-page view, a video view.
• Someone who says “don’t bother under $3,000 a month” ran lead-form or purchase campaigns in a normal-cost industry. At a $27.66 median cost per lead, they are right, and everyone calling them a gatekeeper is wrong.
• Someone who says “$15,000 a month” sells to dentists, where the median cost per lead is $76.71.

None of them are lying. All of them are generalising from one optimisation event to the whole platform. Once you see that, “how much do Facebook ads cost?” becomes an unanswerable question, and “how much does this result cost, and how many do I need?” becomes an easy one.

What Facebook actually costs in 2026

Here is the dataset behind the medians above, with the year-on-year movement, so you can position yourself rather than guess.

Traffic campaigns (optimising for link clicks)

Metric	2025	2024	Change
Median CTR	1.71%	1.57%	+8.2%
Median CPC	$0.70	$0.77	−6.7%

Click costs on traffic campaigns actually fell, and click-through rates rose in 11 of 22 industries. Cheap traffic is still available on Facebook. That is worth knowing, because it is the opposite of the received wisdom.

Lead campaigns (optimising for form submissions)

Metric	2025	2024	Change
Median CTR	2.59%	2.58%	~0%
Median CPC	$1.92	$1.88	+2.1%
Median conversion rate	7.72%	8.67%	−11.0%
Median cost per lead	$27.66	$22.87	+20.9%

This is the number that should shape your planning. Cost per lead rose almost 21% in a single year, and the cause is visible in the row above it: click costs barely moved, but conversion rates fell 11%. People are still clicking. Fewer of them are filling in the form.

Cost per lead by industry

Industry	Cost per lead	Median CPC	Conversion rate
Dentists & dental services	$76.71	$9.78	6.38%
Health & fitness	$52.98	$2.64	5.63%
Beauty & personal care	$51.42	$3.06	5.29%
Home & home improvement	$41.26	$2.23	5.22%
Furniture	$40.04	$2.18	3.77%
Industrial & commercial	$37.34	$1.80	9.34%
Personal services	$30.57	$2.08	6.51%
Education & instruction	$28.22	$1.65	10.08%
Sports & recreation	$19.30	$1.07	5.48%
Arts & entertainment	$18.17	$1.08	9.34%
Attorneys & legal services	$18.17	$4.10	10.53%
Career & employment	$17.64	$0.86	5.77%
Real estate	$16.61	$1.57	9.53%
Restaurants & food	$3.16	$0.74	18.25%

Find your row, multiply by 50, divide by 7. That is your learning floor, and it is a far more useful starting point than any range.

For context: what the same lead costs elsewhere

Facebook looks expensive until you price the alternative. LocaliQ’s 2026 search advertising benchmarks, drawn from thousands of Google and Microsoft Ads campaigns and last updated 1 June 2026, put the search averages at $5.42 per click and $66.69 per lead. Facebook’s $27.66 is well under half of that. Facebook is not the expensive channel. It is the channel where the volume you need before it works is high.

Are Facebook ads getting more expensive? Meta’s own numbers say yes

You do not need a third-party study for the direction of travel — Meta reports it every quarter.

Quarter	Ad impressions delivered	Average price per ad
Q1 2026	+19% YoY	+12% YoY
Q2 2026	+14% YoY	+12% YoY

Sources: Meta Q1 2026 results and Meta Q2 2026 results. Q2 revenue was $60.80 billion, up 28% year over year.

Read those two columns together, because the combination is the interesting part. Meta is increasing supply — 14% more ad impressions delivered — and prices are rising 12% anyway. When supply grows and price grows at the same time, it is not scarcity. It is demand: more advertisers, bidding harder, for the same attention.

Plan for roughly 10–12% annual cost inflation on Facebook and you will not be surprised. If your model only works at today’s prices, it does not work.

The budget mechanics people get wrong

Facebook will overspend your daily budget, on purpose. Meta can spend up to 75% more than your daily limit on a given day to capture an opportunity, then balance it out so that across seven days you never exceed seven times the daily number. A $20/day budget can bill $35 on a good Tuesday. That is the system working, not a billing error — but if you are running on a card with a low limit, budget for the daily spike, not the daily average.

Editing the budget restarts learning. Budget changes count as significant edits, which reset the learning phase and discard the accumulated events. The most common way small advertisers stay permanently learning-limited is by nervously adjusting the budget every two days. Pick the number, then leave it alone for the full window.

Lifetime budgets are not a way to spend less. A lifetime budget spreads the same money across the schedule. If the daily average it implies is below your learning floor, you have the identical problem with a different-looking number.

More ad sets is usually worse at small budgets. Meta lists high auction overlap and running too many ads at once among the causes of learning-limited status. Five ad sets at $10 each is five learning-limited ad sets. One ad set at $50 is one that might actually learn. At small budgets, consolidate.

Three worked budgets

The freelancer with €500 a month. €500 a month is about €16/day. At a $27.66 median cost per lead you would generate roughly 18 leads a month, and the ad set would sit learning-limited the entire time — you would be paying full price for unoptimised delivery. The correct move at this budget is not a smaller lead campaign; it is a different campaign. Optimise for link clicks or landing-page views at $0.70–$1.50 each, capture emails on your own page, and let your follow-up do the conversion work Meta’s algorithm cannot afford to do for you. You are buying traffic and audience data, and you should say so out loud rather than being quietly disappointed each month.

The local service business with €3,000 a month. €3,000 a month is €100/day. In home improvement, at $41.26 per lead, that is 2.4 leads a day — 17 a week. Below the 50-event threshold, but close enough that Meta will optimise reasonably within a couple of weeks. This is the budget band where Facebook lead campaigns genuinely start behaving. Run one campaign, one ad set, three creatives, and do not touch it for fourteen days.

The e-commerce store with €6,000 a month. €200/day clears the learning floor for almost any purchase-optimised campaign under a $28 cost per acquisition, and leaves room to run a second ad set for testing. This is where the “just spend more, it gets cheaper” advice becomes true — not because Meta rewards big spenders, but because you finally cross the volume threshold where its optimisation is working with real data instead of noise.

Five ways the first €500 usually disappears

1. Optimising for a rare event on a small budget. The single most expensive mistake, and the reason for this entire article.
2. Judging a campaign in 48 hours. The first days are learning-phase noise by definition. Meta suggests at least six days; two weeks is the honest minimum.
3. Splitting the budget across five ad sets so none of them ever learn, and they bid against each other in the auction on top of it.
4. Sending traffic to a slow or vague landing page. Conversion rates fell 11% year over year across the board; that decline lands on your page, not on Meta’s. A weak page raises your cost per lead more reliably than any targeting change lowers it.
5. No conversion tracking. Without the Conversions API or a working pixel, Meta optimises toward whatever it can see, which is usually clicks — so you pay lead-campaign prices for traffic-campaign outcomes.

What €30 a day buys you — and the honest comparison

Suppose you have €30 a day, or roughly €900 a month, which is a common freelancer or small-agency number.

On Facebook, optimised for traffic at the 2026 median, that is about 43 clicks a day, 1,300 a month. At a generous 3% landing-page conversion rate you would get around 39 enquiries a month, at roughly €23 each. That is a real, working outcome, and for a business with a proven offer and a good landing page it can be a bargain.

Here is the part the ad-budget articles never say: €900 a month is also roughly what an agency-grade version of the opposite approach costs, and the opposite approach has no cost-per-lead at all.

Every day, in the Facebook groups you already belong to, people post the thing you are paying Meta to help you find: “can anyone recommend a web designer”, “looking for a photographer for a wedding in June”, “our current agency is not working out”. Those posts are already in front of you. They cost nothing to reach. The only reason they do not turn into clients is that nobody can read every group every day, and by the time you scroll past the thread it has 40 comments and the buyer has chosen.

That is the problem ClientRadar exists to solve — it watches the groups, subreddits and feeds you are already a member of, scores each post for buying intent, and drafts a reply in your voice for you to approve. It runs in your own logged-in browser session, so there is no bot logging in as you and nothing gets posted without your tap. We wrote up the method itself, tool or no tool, in the 2026 ban-safe client-finding playbook and specifically for this channel in how to find clients in Facebook groups.

To be fair to ads, because a comparison that only flatters one side is useless: paid beats organic presence on three dimensions. Ads scale on demand — you can triple volume tomorrow, which no amount of community presence can do. Ads reach people who are not in your groups. And ads are predictable in a way conversations never are. If you have a proven offer, a working landing page and the budget to clear the learning floor, Facebook ads are one of the cheapest customer-acquisition channels that has ever existed.

What ads cannot do is work below the learning floor. And that is precisely where most freelancers, consultants and small agencies live. If your realistic monthly ad budget is under about €1,500, the arithmetic in this article is telling you something specific and worth listening to: at that level you are not buying optimised delivery, you are buying a lottery ticket at full retail price. Being genuinely present where your buyers already ask is not a consolation prize at that budget — it is the higher-return option, and the data on where freelance clients actually come from backs that up hard.

The budget-setting procedure, in short

1. Pick the optimisation event you want Meta to chase.
2. Find your target cost for that event — your own data first, the tables above second.
3. Multiply by 50, divide by 7. That is your learning floor.
4. If you cannot afford the floor, change the event — do not shrink the campaign.
5. Commit to at least 14 days before judging.
6. Set it, then stop editing it.

If you would rather not do the arithmetic by hand, our free cost per lead calculator runs it against 2026 channel benchmarks in the browser — no sign-up, and nothing leaves your machine.

Budgets on the other three platforms

Every platform has its own version of these three floors, and the gap between them differs enormously:

• Reddit ads budget 2026 — a $5/day billing floor and the cheapest impressions of the four, with a culture that punishes bad ads in the comments.
• X ads budget 2026 — the only one of the four with no minimum spend at all, and the one where what you pay for is unusually specific.
• LinkedIn ads budget 2026 — a hard $10/day floor, by far the most expensive clicks, and the best measured B2B return on ad spend.

And if the arithmetic above pushed you toward the unpaid side of the ledger, Facebook groups lead generation is the channel page, and social listening for lead generation is the full method.