Facebook Ads Budget Calculator
Sizes a budget against the learning phase: how many ad sets it can keep out of learning, and whether the test split buys enough conversions to decide.
Monthly budget $9,000.00
a day $296.05
Target cost a conversion $28.00
Conversions the budget buys 321.4 a month at that cost
The learning phase floor
conversions an ad set needs 50 a week
which costs $200.00 a day, an ad set
ad sets this budget supports 1
ad sets you have 4
budget each $74.01
conversions each 18.5 a week
verdict consolidate to 1 ad set: at 4 they each sit below the threshold and stay in learning
Testing
test budget $1,800.00, 20% of the total
scaling budget $7,200.00
variants 2
conversions a variant 32.1
a useful test needs about 100 a variant, which is $5,600.00
verdict too thin: at 32.1 conversions a variant only an enormous difference would show, and a smaller one will look like noise
What the budget would need to be
for 4 ad sets out of learning $24,320.00 a month
for a test at this width $5,600.00 in the testing budget alone
If the cost a conversion is wrong
at $21.00 428.6 conversions, floor $150.00 a day an ad set
at $28.00 321.4 conversions, floor $200.00 a day an ad set
at $42.00 214.3 conversions, floor $300.00 a day an ad set
at $56.00 160.7 conversions, floor $400.00 a day an ad set
One ad set needs about 50 conversions a week to leave the learning
phase, which at $28.00 a conversion is $200.00 a day. This budget
supports 1 ad set at that rate, and it is spread across 4.
An ad set stuck in learning does not simply take longer. It delivers at
a higher cost for as long as it stays there, and every significant edit
restarts the phase, which is why frequent optimisation of a small budget
is worse than leaving it alone.
A test needs conversions rather than impressions. At $28.00 a
conversion, 2 variants at 100 conversions each costs $5,600.00 before
anything is learned, and running it on less produces a winner that does
not repeat.
The 20 percent testing split is a convention rather than a finding. What
matters is whether the testing budget is large enough to resolve a
difference at all: below that, the money is better spent on delivery,
and the test is better run later at a size that can answer it.
Consolidation usually beats segmentation on modern delivery. Four ad
sets splitting a budget four ways all learn slowly on a quarter of the
data; one ad set with the same money learns once, and the system finds
the segments faster than the account structure can define them.
Cost a conversion is not stable. It rises with frequency, with
competition and through the fourth quarter, so a budget planned at
today's figure is planning at the best case. The table above shows what
a 50 percent rise does to both the volume and the floor.
None of this is a promise about results. It sizes the budget so the
platform can do its job, which is a precondition for results rather than
a cause of them: the creative and the offer still decide the outcome.
Output is valid and updates as you type.
Fix the highlighted fields to update the output.
Most ad budget calculators multiply a target by a cost and hand back a number. The part that decides whether the plan works is structural, and it is arithmetic you can do in one line: Meta wants roughly 50 conversions per ad set per week before an ad set leaves the learning phase, so one ad set needs 50 times your target cost per acquisition, divided by seven, every day.
At a $28 target that is $200 a day. A $9,000 month is $296 a day, which supports one ad set at that rate and is usually asked to support four.
An ad set that never leaves learning does not simply take longer to settle. It delivers at a higher cost for as long as it stays there, and every meaningful edit restarts the phase.
How to use
- Put in the monthly budget and what a conversion is worth paying for.
- Put in how many ad sets you plan to run. The tool says how many the budget can keep out of learning.
- Set the testing share and the number of variants to see whether the test can buy enough conversions to separate them.
Example
$9,000 a month at a $28 target cost, split across four ad sets with 20 percent set aside for testing:
Monthly budget $9,000.00
a day $296.05
Conversions the budget buys 321.4 a month at that cost
The learning phase floor
conversions an ad set needs 50 a week
which costs $200.00 a day, an ad set
ad sets this budget supports 1
ad sets you have 4
budget each $74.01
conversions each 18.5 a week
verdict consolidate to 1 ad set: at 4 they each sit below the threshold and stay in learning
Testing
test budget $1,800.00, 20% of the total
variants 2
conversions a variant 32.1
a useful test needs about 100 a variant, which is $5,600.00
verdict too thin: at 32.1 conversions a variant only an enormous difference would show
What the budget would need to be
for 4 ad sets out of learning $24,320.00 a month
for a test at this width $5,600.00 in the testing budget alone
If the cost a conversion is wrong
at $21.00 428.6 conversions, floor $150.00 a day an ad set
at $42.00 214.3 conversions, floor $300.00 a day an ad setTwo decisions fall out of that. Consolidate to one ad set, and stop running the test until there is enough money to answer it.
Pitfalls
The floor scales with your target cost. A $60 target needs $428 a day an ad set. Expensive conversions need much larger budgets to optimise at all, which is why a high-value product with a small budget should usually optimise for an earlier event in the funnel.
Splitting a budget across ad sets divides the learning, not just the money. Four ad sets each learn from a quarter of the data. One ad set learns from all of it and the delivery system finds the segments faster than an account structure can define them.
Editing restarts the phase. Budget changes above about 20 percent, new creative, changed targeting and a new optimisation event all reset it. Daily optimisation of a small account keeps it permanently in the worst-performing state.
A test needs conversions, not impressions. Around 100 conversions per variant is the rule of thumb for seeing an ordinary difference. Below that a winner is mostly noise, and the winner will not repeat when it is scaled.
The 20 percent testing split is a convention. What matters is whether the split is large enough to resolve anything. If it is not, the money does more good in delivery and the test is better run later at a size that can answer it.
Target cost is not stable. It rises with frequency, with competition and through the fourth quarter. Planning at today’s figure is planning at the best case, which is what the sensitivity table is for.
Sizing the budget correctly does not make the ads work. It removes one specific obstacle. The creative and the offer still decide the outcome.
Compatibility
Arithmetic in the browser: nothing is uploaded and nothing is stored.
The floor is the weekly optimisation-event threshold multiplied by the target cost and divided by seven, and the threshold is editable because some objectives are cheaper and some accounts optimise for an earlier event. Setting it to 25 halves the floor and the number of supported ad sets doubles, which the test suite asserts.
Monthly figures use 30.4 days, so a daily budget converts back to the same monthly total rather than drifting by a day and a half each month.
The three verdicts are distinct cases: every ad set above the threshold, too many ad sets for a budget that could support fewer, and a budget below the floor for even one ad set, where consolidating cannot help and the target cost or the budget has to change.