If you already run Meta ads, you have probably watched an ad set sit in "Learning" or slide into "Learning Limited" and wondered whether the fix is more budget. Usually it is — but the right number is a calculation, not a round figure someone posted in a Facebook group. This page gives you the formula, works it against real operating numbers, and then does the part the budget guides skip: whether that spend can actually turn a profit.
The formula for your learning-phase minimum
The minimum budget to clear the learning phase is set by one input you control — your target cost per purchase (CPA) — and one threshold Meta sets. The math:
minimum daily budget per ad set = (target CPA × 50) ÷ 7
So if you aim to acquire a customer for $18, your minimum is 18 × 50 ÷ 7 = about $129 per day, in a single ad set. For a $25 target CPA it is 25 × 50 ÷ 7 = about $179 a day. The number scales with how expensive your conversions are, which is why a low-CPA lead campaign and a high-AOV purchase campaign have wildly different "minimums."
This is the same formula the practitioner guides land on. One Facebook ads budget guide writes it as "Target Cost Per Lead x 50 / 7 days = Minimum daily budget," and Jon Loomer frames it as multiplying your projected cost per action by fifty for the week, then dividing by seven for the day in his budget breakdown.
Why Meta needs about fifty events a week
The fifty in that formula is not arbitrary. An ad set exits the learning phase once it gathers roughly fifty optimization events — whatever you optimized for, usually purchases — inside a rolling seven-day window, a benchmark that same budget guide attributes to Meta's own best practices.
Below that rate, delivery stays unstable and your cost per result is higher and more volatile, because the system is still exploring who to show the ad to. Starve it long enough and the ad set lands in "Learning Limited" — a status meaning it will probably never gather enough events at its current budget and audience to stabilize.
So the budget question is really a conversion-rate question in disguise. You are not buying a magic number; you are buying enough weekly conversions to let the algorithm settle.
A worked example on real operating spend
Say your store does 340 orders a month at a $31 average order value, and you run about $2,800 a month in Meta spend. That is roughly $93 a day across the whole account.
Now set a target CPA of $18. At $93 a day you buy about 5.2 purchases a day, or around 36 a week — under the fifty threshold. If that $93 is split across three ad sets, each one sees only about twelve events a week and none of them ever exits learning. You are paying the learning tax three times over.
To clear fifty events a week at an $18 CPA you would need 18 × 50 ÷ 7 = about $129 a day in one ad set — roughly $3,860 a month. Your account cannot feed three ad sets that volume. The fix is not always more money; often it is consolidating budget into fewer ad sets so each one actually reaches the threshold. For the deeper mechanics of how long this takes and what resets it, see our guide on how long the Facebook ads learning phase lasts.
Why a dollar a day keeps you stuck
Meta's technical floor is low — about a dollar a day for impression and click objectives and roughly five dollars a day for conversion objectives, per that minimum-budget guide. Those numbers let the ad serve. They do not let it learn.
At five dollars a day with an $18 CPA you earn about two conversions a week — the ad set never leaves learning and your cost per result stays stuck at its worst, most volatile level. Running a one-dollar-a-day conversion ad set and expecting optimization is the single most common budget mistake. We walk through where the dollar-a-day idea actually applies, and where it falls apart, in the one-dollar-a-day Meta ads budget breakdown.
The same guide's rule-of-thumb minimums by goal reflect this gap: it suggests roughly $50–$100 a day for conversion and sales campaigns, versus $10–$20 a day for awareness — because sales objectives need far more events to stabilize. Treat those as floors to even bother, not targets.
The number the budget guides skip: break-even
Here is what almost every "minimum budget" article leaves out. Hitting fifty events a week only matters if those events make money. The learning-phase budget and your profit are the same conversation.
Your break-even ROAS is simply 1 ÷ your contribution margin — the share of revenue left after product cost, shipping, and fees, before ad spend. At a 45% contribution margin, break-even ROAS is 1 ÷ 0.45 = about 2.22x. On a $31 order that 45% is $13.95 of gross profit, which is the most you can pay to acquire that order before the sale loses money.
Notice the collision: the learning-phase math wanted an $18 target CPA, but the margin math says you can only afford about $14. Feeding the algorithm fifty events a week at a CPA your margin cannot cover just buys fifty unprofitable orders faster. That is why Jon Loomer's own budget breakdown notes the formula can imply $10,000-plus a month and simply "isn't realistic for many small businesses" — the spend to satisfy the algorithm can outrun the economics.
The real move is often to raise average order value so your affordable CPA rises to meet the learning-phase requirement — a bundle or a post-purchase upsell that lifts margin dollars per order does exactly that. See the cluster hub on profitable ad scaling for how AOV work buys you headroom to spend.
Fewer ad sets, more events
If your total budget can only produce, say, forty conversions a week, splitting it across four ad sets guarantees all four stay in learning. Pointing it at one ad set gets you to the threshold. Consolidation is usually cheaper than adding budget.
The tradeoff: fewer ad sets mean fewer clean reads on distinct audiences or creatives. Early on, use separate ad sets only for genuinely different tests, and collapse them once you know the winner. For the step-by-step cadence of raising budgets without re-triggering learning, our budget-scaling best practices covers the safe increments.
Check your tracking before you raise the budget
The fifty-event count is what Meta sees, not what actually happened. If your pixel or Conversions API drops events — iOS privacy, ad blockers, a broken tag after a site deploy — Meta undercounts your conversions and the ad set stays trapped in learning even though real-world sales were fine.
So before you throw more money at a stuck ad set, reconcile Meta-reported purchases against your actual store orders for the same window. If your backend shows fifty sales and Meta shows twenty, the problem is measurement, not budget. Pouring spend into a tracking leak is how accounts bleed cash while staying "in learning" forever. If lifetime-budget scheduling is part of your setup, the ad scheduling and lifetime budget guide covers how pacing interacts with event volume.
How Victor sizes the budget for you
Working out the real learning-phase minimum means holding three numbers together: your target CPA, your true per-order margin, and the events each ad set is actually clearing. Most sellers only ever see the first.
Victor is the AI employee inside PodVector AI. He computes your true per-order profit by seeing every order, every ad dollar, and every supplier cost from Shopify, Meta Ads, Printify, Printful, and Gelato together — so the affordable CPA he works from is real, not a guess. He changes budgets and pauses ad sets directly in Meta Ads, consolidates spend so ad sets actually reach the event threshold, and tells you when a "minimum budget" would just buy unprofitable orders.
That turns the formula on this page into a decision Victor acts on, not a spreadsheet you maintain. Put Victor on your ad account and let him size the budget against your real margins.
FAQs
What is the absolute minimum budget to run Facebook ads?
About a dollar a day for impression, click, and video objectives, and roughly five dollars a day for conversion objectives, according to this minimum-budget guide. But that is the floor to serve an ad. To actually exit the learning phase you need enough budget to clear about fifty conversions a week per ad set, which is almost always far more.
How do I calculate my learning-phase budget?
Multiply your target cost per purchase by about fifty, then divide by seven. That is your minimum daily budget per ad set. A $20 target CPA works out to 20 × 50 ÷ 7 = about $143 a day in a single ad set.
Why is my ad set stuck in "Learning Limited"?
Usually one of three things: the budget is too low to hit fifty events a week, too many ad sets are splitting the same conversions, or your pixel and Conversions API are undercounting events that really happened. Check tracking first, then consolidate ad sets, then consider more budget.
Should I raise the budget to escape the learning phase?
Only if the extra spend buys events at a cost per purchase your margin can cover. Your break-even is 1 ÷ your contribution margin, and if the learning-phase CPA sits above what you can afford, raising budget just buys losing orders faster. Often it is better to consolidate ad sets or raise average order value instead.
Does changing the budget reset the learning phase?
Large budget changes count as a significant edit and can re-trigger learning; small nudges generally do not. There is no official Meta percentage for "safe," but keeping increases modest and spaced out is the common practitioner default — the budget-scaling best practices guide covers the increments.
Is fifty events a week a hard Meta rule?
It is Meta's published benchmark for exiting learning, cited in this budget guide, but it is the count Meta sees. Accounts sometimes stabilize with fewer events, and tracking gaps can make a healthy ad set look short. Treat fifty as the target to plan budget around, not a line in the sand.