TL;DR
- There is no right budget, there is a minimum. Ad platforms need about 50 conversions a week to finish learning who to target, so your minimum daily spend is target cost per conversion times 50, divided by 7.
- A $30 lead means about $214 a day, not $5. The '$5 a day' answer is from 2018, before the algorithm's data needs set the minimum instead of your wallet.
- Spend below that minimum and you are not running a cautious test. The platform never gets enough data to stop guessing, and your weekly numbers swing for no reason.
- Three times the honest answer is 'not ads yet.' No converting offer, revenue near zero, or no working tracking. A small budget on a broken funnel just burns slower.
- Before you spend a dollar, write three numbers down. Your minimum daily spend, your total test budget, and the cost per customer at which you shut it off. The budget buys an answer, not just traffic, so decide what the answer is worth before you start.
"How much do I need" is the wrong question, and you already know it
The question you type is not really about a number. It stands in for a fear. How much will I waste before I know whether this works?
What you really want is two things. Permission to start without lighting money on fire, and a way to know when to pull the plug.
The versions of the question I hear on mentor calls are barely disguised. "Is 3K a month enough?" "What should my daily budget be, and how much is a fair spend per conversion?" And the one that contains this whole post, "If I need 50 purchase events per week to exit the learning phase, how can I ever make Meta work at $30 a day?"
The number usually arrives with an apology attached. One founder, asked for her budget directly, "Oh God. Honestly, we have an insanely tiny budget." Another spent £2,000 over two months for two or three leads, and only heard the words small budget from the mentor afterward, because from the inside it felt like serious money.
Nobody should be embarrassed by arithmetic no one ever showed them.
The algorithm sets your minimum budget, not you
Below a certain daily spend, the platform never stops guessing.
Modern ad platforms run on machine learning that needs a minimum number of conversions each week to figure out who to show your ads to. Until it has seen enough, it sits in what the platforms call the learning phase, which is a polite name for guessing. Pick a budget below that minimum and you have not bought a smaller test. You have bought more guessing.
The tell is a campaign that looks busy and teaches you nothing. Impressions, clicks, a little spend every day, and at the end of the month you still cannot say whether it works.
I paid for this lesson at EuroVPS, my first marketing job, running AdWords at $10 a day with more confidence than sense and wondering why the reports told me nothing month after month. The budget was too small to learn from, let alone win with.
This is why the old answer died. "Five dollars a day" made sense when you set targeting by hand. On a conversion goal in 2026 the minimum is set by the platform's AI, not by your wallet, and it is higher than you want it to be.
The wrong mental model produces scenes like this, a jewelry brand running €5 a day for its first five days "as a learning phase," then bumping to ten. The learning phase is not a warm-up you schedule. It ends only when the platform has seen enough conversions, and at €5 a day it never ends.
So the number comes from arithmetic, not taste. Here it is.
your minimum, as arithmetic
The 50 is the conversions the algorithm needs each week to stop guessing. So target CPA times 50 is one week of spend. Divide by 7 for the daily minimum.
Not what you can afford. What the algorithm requires. If the minimum is out of reach, that is a real answer too, and a cheaper one to get now than in a month.
One trap in the math, because it is where people fool themselves. The 50 is weekly, so target CPA times 50 is a single week of spend. Divide by 7 for the daily figure, and if you want a monthly number, multiply the daily one by 30. A $30 lead is roughly $214 a day and about $6,400 a month, and that lead cost is not hypothetical. Average lead-gen cost per lead ran near $28 in 2026, up about 20% in a year.
Fifty is not a number someone picked to be annoying. It is roughly how many examples the bidding model needs before its guesses about who will convert beat a coin flip. Under that, it is extrapolating from too little data, which is the polite way to say guessing with your money.
One honesty note. Meta documents the 50-a-week threshold, Google's version floats closer to ten conversions a day, and mentors quote both depending on the platform in front of them. Treat 50 a week as the planning number, not physics.
And the newer campaign types raised the minimum, they did not lower it. Google recommends a daily budget of at least three times your target cost per action for Performance Max, and Advantage+ Shopping guidance lands around $100 a day. Automation is hungrier than manual ever was.
When this math finally lands on a call, the reaction is usually relief, not despair. "That really clears things up for me. That's something I was worrying about a lot." A rule you can compute beats a fear you can't.
How to tell if your budget is too small
Under-funded does not mean cautious. It means paying for noise.
The gut-check is one question. Can each ad set plausibly get one conversion a day? Ten dollars a day against a $100 cost per conversion cannot, so the platform never sees enough conversions to finish learning, and the weekly result swings for reasons you will never find. Thirty dollars a day at a $70 cost per conversion, split across two ad sets, has the same problem, too few conversions per week for the automated bidding to learn anything.
One founder ran exactly that math live on a call, ten dollars a day against a hundred-dollar target, his own account on the screen.
"Oh shit."
Run that arithmetic on your own case before you launch. Divide your daily budget by the number of ad sets, then by your cost per conversion. If the answer is less than one conversion a day per ad set, you already know the campaign will never finish learning.
You found that out for free, before spending a cent.
I watched a mobile app pour about $2,000 a month into install campaigns that restarted their learning phase every three or four days, because every small edit reset a learning process that never had enough conversions to finish. Months of spend, no readable answer, because the account never held still long enough to gather the data a conclusion needs.
The instinct when a small budget underperforms is to spread it, a brand campaign and a competitor campaign and a marketplace campaign, a dozen ad sets, twenty creatives. That is backward. Each split divides your already-few conversions across more campaigns, so none of them reaches the minimum and all of them keep guessing.
the same $30, two ways
The same money buys one readable test or five that never leave the learning phase. On a small budget, splitting is how you burn it.
So if the budget is fixed and below the minimum, the move with the most payoff is changing what you optimize toward, not adding money. Move the conversion event one step up the funnel, from purchase to add-to-cart, from a signed contract to a started trial, and pick the deepest event that will still happen about 50 times a week. Just do not go so shallow that you optimize toward raw page views, or you train the machine to buy junk.
On Google specifically, a small budget survives by staying deliberate. Start on manual cost-per-click so the account cannot outspend your logic, aim at the long-tail, low-competition keywords the big players ignore, and build an aggressive negative-keyword list so you stop paying for searches that will never buy. Track a cheaper micro-conversion, reaching the checkout or a pricing-page visit, as a rough five-dollar stand-in while real purchases are still too rare to give the bidding anything to learn from. Hand the reins to Maximize Conversions only once you are past the 50-a-week minimum.
And cap what Performance Max can spend.
Left uncapped, Performance Max will eat your Search budget within a couple of weeks, because it chases the cheapest conversions it can find, and cheap rarely means valuable. On a small budget that is exactly the money you needed for high-intent clicks.
Three times the honest answer is "not ads yet"
A perfect budget on a broken funnel still burns. Before you compute a minimum at all, three situations mean the answer is nothing, yet.
- No converting offer yet
- If nothing sells through conversations, outreach, or organic, ads only amplify a zero. Ads scale a working offer, they do not discover one. Fix the offer first, and if that is the real gap, start with your ICP and positioning.
- Revenue near zero
- Around a thousand a month in, the budget that would move the needle is bigger than the business can feed. Grow the base with unscalable channels first, then amplify the winner with spend.
- No working conversion tracking
- If the platform cannot see a real sale, it optimizes toward noise, and you cannot read the result anyway. Get tracking firing before a dollar goes in. Our guide to auditing your own ad account walks that exact check.
Under all three sits a ceiling nobody can raise with budget, what a customer is worth. If your margin-adjusted room to acquire a customer is $40, no daily number makes a $120 acquisition cost profitable. The budget question has a hard cap, and the cap is set by the business, not the ad account.
The internet is full of the receipt for skipping this. Whole writeups exist with titles like we wasted $50K on Google Ads so you do not have to, and the pattern under almost all of them is the same, a real budget poured onto an offer nobody had proven anywhere cheaper first.
The counter-move costs almost nothing. Plenty of founders prove or kill a concept for a couple hundred dollars, sometimes for zero, through conversations and outreach, before a cent touches an ad account. Ads are an amplifier, and an amplifier does nothing for a signal that is not there yet.
Write your kill-number down before you spend a dollar
This is the move that separates a test from a donation.
It is also where the page-one budget guides go missing. Madgicx, WebFX, the whole calculator set will hand you the same formula I just did and stop, as if the only question were arithmetic. None of them asks whether you should be spending at all yet, and none of them makes you write down where you will stop. Decide in advance what you will spend to buy one clean yes or no, and put the stopping point on paper, the kill-number, the cost per customer at which you stop.
written before dollar one
Minimum: $214/day, from a $30 lead1
Test budget: ~$6,000, the minimum across 4 weeks2
Kill number: stop above $45 cost per customer3
The minimum
What the algorithm needs to stop guessing. Below it, nothing you read is real.
The tuition
The daily minimum times the weeks to a verdict. This is what the decision costs, in full.
The kill number
The cost per customer at which you stop, chosen now, while you are still calm.
Three numbers on one line before launch. It is the difference between a test and a donation.
Then hold the window. Give the account two to three months and expect the first couple of weeks to be pure learning, where the numbers mean almost nothing. The budget is buying you one answer, does this channel work for us. The kill-number is what stops you paying for that answer twice.
How-to guide
Before you spend a dollar
The whole post as five moves, in order. Do them before the account is even funded.
Compute your minimum
Target cost per conversion times 50, divided by 7. That is your minimum daily budget, the spend the algorithm needs to stop guessing. If it is out of reach, that is your answer for now.
Check you are ready
A converting offer, revenue to feed the spend, and conversion tracking that actually fires. Miss one and the honest answer is not ads yet.
Pick one thing to optimize toward
The deepest event you can still feed about 50 times a week. Purchase if you can reach it, add-to-cart or a qualified lead if you cannot.
Concentrate, do not spread
One objective, one or two ad sets, your best three creatives. Splitting a small budget starves every piece of it.
Write the kill-number down
Your minimum daily spend, your total test budget, and the cost per customer at which you stop. Set it now, while you are calm, before the first dollar moves.
%20(1).png)
Budget too small, or offer broken?
Book a 1:1 call with a paid buyer who has scaled from small budgets and screen-share your account. Bring your cost per conversion and your daily spend, leave knowing whether to raise the budget or fix the offer. One membership, unlimited calls.
The formula is the easy half
An assistant hands you the formula in ten seconds. The hard half is everything the formula assumes.
Whether your $30 target cost per lead is even realistic in your market. Whether the offer is the real leak and no budget will fix it. Whether you are one setting away from working or a month away from an expensive nothing. Those answers do not live in a formula. They live in your actual account, read by someone who has opened a few hundred of them.
Someone who has read that many can tell in ten minutes whether your $30 target is a fantasy for your category, whether the tracking is feeding the model at all, and whether you are one budget change from working or a rebuild away. That read is the half the formula cannot give you.
That is what GrowthMentor is for. One membership is unlimited 1:1 calls with vetted paid-media operators who have scaled from small budgets, and if you would rather not go picking, you post the question and the ones who run Google and Meta all day raise a hand. Here is that exact request, three hands up.
The kill-number keeps you honest. A good operator tells you where to point it.
Suggested mentors
A few of the paid-media operators people book to pressure-test a budget:
Luis Camacho
Scales SaaS with paid ads across Google and Meta. Knows where budget moves pipeline, and when it can't.
Agata
Ex-Google, six years on the Ads team. Makes Google Ads work for SaaS and B2B, tracking built for a considered sales cycle.
Nilay Jayswal
Fractional performance marketer across Google, Meta and LinkedIn. Squeezes signal out of a small budget instead of adding money.
%20(1).png){kind=small}
Sam Collier
Facebook and Instagram ads with $350M+ in ad-driven revenue behind him, plus landing-page CRO. Reads the funnel, not just the budget.
The full live roster is at mentors for Google Ads and mentors for Meta and Facebook ads. And once you have decided to spend, the next thing worth reading is how to audit the account yourself before you hand it to anyone.
Ad budget FAQ
Vetted paid-media operators, every one included
Before you fund the next month.
Find out if it's the budget, or the offer.
Book a 1:1 call with a vetted paid-media operator, screen-share your account, and compute the minimum together. One membership, unlimited calls, every mentor included. No per-session fee, no pitch.
Talk to a mentorKeep reading
More from the GrowthMentor blog
Paid Ads · Jun 19, 2026
How to Audit Your Own Ad Account Before You Pay Anyone To
Foti Panagiotakopoulos
Marketing · May 29, 2026
Why Your Ads Get Clicks but No Conversions
Foti Panagiotakopoulos
Marketing · Jun 07, 2026
How to Define an ICP You Can Actually Use
Foti Panagiotakopoulos
Paid Ads · Apr 08, 2026
How to find a Facebook ads mentor. And why it beats another course.
Foti Panagiotakopoulos
