How Much to Spend on Facebook Ads (2027)

There is no fixed price, only a budget you can hold

The honest answer to "how much should I spend on Facebook ads" is that there is no fixed price, only a number you can defend for long enough to learn something. Meta does not sell ad space at a sticker rate. You set a daily or lifetime budget, and an auction prices each result against your competitors, your audience, and your creative. That is why two stores in the same niche can pay $4 and $40 for the same kind of sale: the budget field is yours to set, the cost per result is not.

So the useful question is not "what is the right number" but "what number lets me read a real result, then exit the learning phase, then grow." Those are three different budgets at three different stages, and conflating them is where most spend gets wasted. This guide walks the ladder in order: pick a test budget you can hold steady for about two weeks, raise it to a learning-phase floor that funds roughly 50 optimization events a week so Meta can actually optimize, then scale from that floor once you have a winner. Money here is in USD, but the framework is currency-neutral: every figure below works if you swap in your own currency and your own target cost per result.

One number to carry through all three stages: an ad set exits the learning phase after it gathers about 50 optimization events within roughly 7 days of its last significant edit (Meta Business Help Center, 2026). Almost every budget decision below traces back to that threshold. If you want the full mechanic behind it (what Learning Limited means, which edits reset the clock, the 2026 shift toward adding ads instead of editing them), the Facebook ad learning phase guide covers it. This page borrows only the 50-events number and spends its depth on the money math.

Stage 1: the test budget, sized to buy a verdict

Your real day-one question is not "how much to spend forever," it is "how much do I risk before I have any data at all." The answer is psychological before it is mathematical. Pick a test budget you can hold steady for about two weeks without flinching, because the fastest way to waste money on Facebook is to panic on day two, kill the ad before it has spent a single full cycle, and start over with nothing learned. A budget you cannot hold is worse than a smaller budget you can.

Inside that comfort range, size the test so it can plausibly buy enough results to read a real cost per result, not just impressions. You are trying to buy a verdict: roughly 10 to 30 of whatever action you care about, spread over the test window, so that the cost number you see is a signal and not noise. One or two conversions tell you nothing; the variance is too high to act on. Ten to thirty start to draw a line you can trust.

How much that costs depends entirely on what one result costs in your category, which is why a single starter number cannot fit everyone. WordStream's 2025 benchmarks put the average Facebook cost per lead at $27.66, up about 21% year over year from $22.87 (WordStream/LocaliQ, 2025), but the spread underneath that average is enormous. Lead-objective clicks ran as cheap as $0.74 for restaurants and food and as dear as $9.78 for dentists and dental services in the same dataset (WordStream/LocaliQ, 2025). A test budget that buys 20 leads for a restaurant could buy two for a law firm. So do not copy a number off a blog. Estimate your own target cost per result from your category and your margins, then size the test to clear a readable count of events at that cost.

A quick sanity check on what a test buys: at an average Meta CPM of about $8.19 in 2025 (Gupta Media, 2025), $200 of spend buys roughly 24,000 impressions. Whether those impressions turn into 5 results or 50 is the creative's job, not the budget's, which is exactly why a cheap CPM never guarantees a cheap result. (If your CPM itself looks alarming, the Facebook ad CPM guide explains why it is an auction-priced score you mostly do not control.) The test budget exists to answer one question: at my real CPM and my real creative, what does one result actually cost me? Once you have that number, stop guessing and move to Stage 2, because that number is the input the whole floor formula needs.

Stage 2: the learning-phase floor, the most important number on this page

This is the number most "how much should I spend" guides never give you, and it is the one that decides whether your money does anything at all. Below this floor, Meta's delivery system never collects enough signal to optimize, so it spends your budget inefficiently and your cost per result stays stubbornly high. At or above it, the system can learn, and your costs start to behave.

The 50-events floor, derived

Start from the threshold: an ad set needs about 50 optimization events in 7 days to exit the learning phase (Meta Business Help Center, 2026). Turn that into a daily budget and you get a community heuristic that practitioners lean on:

Daily budget per ad set = (your target cost per result x 50) / 7

It is simple arithmetic. Fifty events times what each one costs you is the weekly spend you need; divide by seven for the daily figure. This is not an official Meta formula, just a sanity check derived from Meta's own 50-events threshold, so treat it as a floor to clear rather than a guarantee. But it stops you from launching an ad set that was mathematically doomed to Learning Limited from its first hour. Here is the floor across a range of common target costs:

Read that table once and the whole budget question reorganizes itself. If your product sells for a $40 cost per purchase, the learning floor is close to $286 a day per ad set, or about $2,000 a week. If that number makes you wince, good: it means you now know your real constraint before you spend, not after. The localizing rule for any currency is even simpler than the formula: multiply your target cost per result by about 7 to get the daily floor (50 divided by 7 is roughly 7). A 30-unit target cost in any currency points at roughly 210 units a day. Plug in your own number and your own currency, and the floor falls out.

Meta's own 5x rule: a second way to the same floor

There is a second, more authoritative anchor that almost no guide pairs with the first. If you use Meta's Cost Per Result Goal bid strategy (where you tell the system the cost per result you are willing to pay), Meta's own guidance is to set your daily budget to at least 5 times that goal (Meta Business Help Center, 2026). A $10 cost-per-result goal points at a $50 daily minimum; a $30 goal points at $150. Set it tighter and you choke delivery before the ad set has a fair chance to find results at your target price.

The two rules are not in conflict, they are two routes to the same conclusion: a serious conversion budget is several multiples of one result's cost, not a flat $5 or $10 a day. The 5x rule is Meta's floor for one bid strategy; the 50-events math is the practitioner floor for exiting learning. When you are unsure, size to the higher of the two and you will not starve the ad set.

Why $5 a day is a different conversation

Meta's technical minimums are genuinely low: about $1 a day for impression-based campaigns and about $5 a day for campaigns optimized for clicks, conversions, or other lower-frequency events (Meta Business Help Center, 2026). Those are the floors that keep a campaign live, not the floors that make it work. A $5-a-day budget optimized for purchases at a $30 cost per purchase buys about one result a week. The learning phase wants fifty. The system never gets its footing, delivery stays erratic, and your cost per result reads high and jumpy. The technical minimum tells you what Meta allows; the formula tells you what your objective actually needs. Mind the gap between them, because that gap is where small budgets quietly burn.

The event you optimize for is a budget dial

Here is the lever that lets a small budget actually work, and it is the most underused idea in budget sizing. The floor formula has two inputs, the 50 events and your cost per result, and you can only change one of them: the cost of the event. So choose a cheaper event, and the whole floor drops.

Watch what happens to the same business at different optimization events:

Same store, same product, three completely different budget floors, because deep, rare events (a purchase) cost more per event and therefore demand a far higher budget to hit fifty of them in a week than cheap, frequent events (an add-to-cart or a lead) do. This is the cash-strapped advertiser's real out. If your purchase floor of $286 a day is out of reach, optimizing for add-to-cart at roughly $36 a day can keep an ad set learning on a budget you can actually hold. You trade some signal quality (an add-to-cart is a weaker buy intent than a purchase) for a budget that clears the threshold. For a small or new account, a funded ad set learning on a cheaper event beats a starved ad set stalled on the perfect one. As volume and confidence grow, you can move the optimization deeper toward the purchase and let the budget rise with it.

Per ad set, not per account: why fragmenting multiplies the floor

The single most common budgeting mistake is conflating "my monthly budget" with "what I set in Ads Manager." They are not the same, and the difference can quietly sink an otherwise healthy budget. The 50-events floor is per ad set, not per campaign and not per account. Every ad set runs its own learning phase and needs its own roughly 50 events a week to exit it.

That has a hard consequence. Split a $214-a-day budget (the floor for a single $30-cost ad set) across five ad sets, and you have funded none of them. Each one now gets about $43 a day, far below its own floor, so all five stall in Learning Limited together while you wonder why a "decent" total budget is performing badly. The signal you needed in one place is shattered into five thin streams, none of which crosses the line.

So the real budget question is not "how much should I spend" but "how few ad sets can I afford to fund properly." Fewer, better-funded ad sets concentrate the signal and exit learning; more ad sets divide it and stall. Practically: consolidate. Lean on Advantage+ Campaign Budget so Meta routes one pooled budget toward whichever ad set is clearing events fastest, rather than guaranteeing every ad set a starvation ration. If your total budget can fund exactly one ad set above its floor, run one. One ad set that learns beats five that never do.

Afford versus need: sizing spend to margin, not to a percentage

Here is the honest tension the formula creates. The learning floor might say $214 a day. Your margins might say $50 a day is the most you can spend and still profit. Both numbers are real, and pretending the conflict away is how accounts lose money. When the budget Meta needs is higher than the budget you can afford, you have exactly four honest moves, and a generic "spend more" is not one of them.

First, optimize for a cheaper event so the floor drops into your range, as in the section above. Second, consolidate ad sets so your affordable budget funds one ad set above its floor instead of several below it. Third, accept a longer, Learning-Limited ramp: run below the ideal floor knowing delivery will be erratic and your read will take longer, which is a defensible choice for a brand-new account finding its feet. Fourth, and most honest of all, do not run a conversion campaign yet. If your margin cannot fund even one ad set to fifty cheap events a week, your money is better spent on a tighter offer or stronger creative first, because budget cannot rescue weak creative: across a large meta-analysis, roughly 49% of a brand's sales lift traced to creative quality (Marketing Charts/NCSolutions, 2024). The ad does the lifting; the budget only buys it impressions.

To know which move is yours, work backwards from margin, not from a revenue percentage. Your break-even cost per result is simply your profit per sale: if a product sells for $60 at a $25 margin, you can pay up to $25 to acquire a sale before the campaign loses money, which sets a break-even ROAS of about 2.4 ($60 revenue / $25 spend). Anything cheaper than $25 per sale is profit. That margin number, not a generic budget rule, is what tells you whether your affordable budget and Meta's needed budget can ever meet, and at which optimization event they meet. Size spend to the profit in a sale, then check it against the learning floor, and the two numbers either reconcile or they tell you to fix the offer first.

The "5 to 15% of revenue" rule, and why it cannot size an ad set

Almost every generic budgeting guide anchors on the same heuristic: spend 5 to 15% of revenue on marketing. It is not wrong, exactly. Gartner's 2025 survey put the average marketing budget at about 7.7% of company revenue (Gartner, 2025), so the range is a real planning benchmark. But it answers a completely different question than the one this page is about, and treating it as a way to size a conversion campaign is a category error.

A revenue percentage is a top-down, whole-company planning figure: it tells a finance team roughly how much total marketing should cost across every channel, brand, content, agency fees, and paid ads combined. It says nothing about whether a single Facebook ad set can buy 50 optimization events in 7 days. You could allocate a textbook 10% of revenue to marketing and still split it so thin across channels and ad sets that no Facebook ad set ever exits learning. The percentage rule sizes a marketing department; the floor formula sizes an ad set. Use the percentage to decide how much money exists for marketing at all, then use the 50-events floor (bottom-up, from your real cost per result) to decide how that money has to be deployed inside Facebook so it actually works. One is the ceiling on your total; the other is the floor under each test. They are not substitutes.

Stage 3: scaling from the floor, without resetting it

Once an ad set is past the learning phase and profitable, the floor stops being a floor and becomes a launchpad. Now the question flips from "is my budget high enough to learn" to "how fast can I add budget without breaking what is working." The trap is obvious in hindsight: pour budget in too fast and you trigger a significant edit, which restarts the learning phase and resets the very ad set you were trying to grow. Large budget jumps are commonly cited as triggering a reset at around 20% or more, though Meta does not always publish an exact figure (Meta Business Help Center, 2026), so the practitioner convention is to raise budget in steps of roughly 20% every few days and let delivery re-stabilize between bumps rather than doubling overnight.

That is the on-ramp; the full climb is its own subject. Vertical scaling (raising the budget on a proven ad set in measured steps) and horizontal scaling (duplicating a winner into new audiences) each have their own mechanics, and rushing either one wastes the spend that got you partway. The detail lives in the guide to scaling a winning Facebook ad; the only thing to carry from this page is that scaling starts from the floor you funded in Stage 2, and it climbs in small, patient steps so the system never has to relearn from zero. And once you are spending real money, the next job is making each dollar go further: the playbook for lowering Facebook ad costs covers the levers (creative refresh, audience overlap, placement mix, bid strategy) that cut cost per result on a budget you already run.

Putting the ladder to work this week

You do not need a perfect budget, you need the right number for the stage you are in. Estimate your target cost per result from your category and your margin (your profit per sale is the ceiling). Pick a test budget you can hold steady for two weeks and that can plausibly buy 10 to 30 results, and resist the urge to kill it early. When you see a real cost per result, run the floor formula, target cost times 50 divided by 7, to find what one ad set needs to exit learning, and cross-check it against Meta's 5x rule if you use the Cost Per Result Goal bid strategy. If that floor is out of reach, pull the event-choice lever or consolidate to one funded ad set before you do anything else. Then, and only then, scale from the floor in 20% steps. Three numbers, in order: a test budget, a learning floor, a scaling step. That is the whole answer to "how much should I spend." For readers whose intent is genuinely local, the Facebook ads cost guide for Malaysia runs the same logic in ringgit with local benchmarks.