Free tool · no sign-up

Creative test calculator: how many ads you can test, and when you'll know

Enter your target cost per purchase and the number of new ads you want to test. See how many days it takes to reach verdicts, how many new ads a month you can test and the lines to judge each one by.
$
= $10.30 per ad
$
optional
$

Test budget

$61.80

6 ads × 2× × $5.15 target cost per purchase

Days to verdicts

about 5 days

$61.80 ÷ $15.00 a day

Ads you can test a month

30

about 7 a week, from 20% of $1,546

Stop line per ad

$10.30

spent with no sale → pause it (2× target, the default Autopilot uses)

Winner signal

$15.45

roughly what three sales at your target cost

Meta doesn’t split budget evenly: some ads will spend more than this, some almost nothing. Treat an ad that barely spent as untested, not as a loser.

How to use the calculator

Enter a few numbers you already know:

  • Currency: pick yours from the list. The maths is the same in any of them.
  • Target cost per purchase: your target CPA — the most you can pay for a sale and still be happy. If you haven't set one, start from your margin on a first order.
  • New ads in the test: how many ads you want verdicts on.
  • Spend per ad before judging: a multiple of your target CPA — 2× by default, anywhere from 1× to 3×.
  • Daily testing budget: what you'll spend a day on the test.
  • Monthly ad budget and testing share (optional): your total monthly spend on Meta, and the part of it you set aside for new ads — 20% by default.

You'll see what the test costs, how many days it takes to reach verdicts and, if you add a monthly budget, how many ads a month you can afford to test. You'll also get the two lines to judge each ad by: when to stop it, and when it's starting to look like a winner.

The formulas and assumptions

Result Formula
Test budget new ads × multiple × target CPA
Days to verdicts test budget ÷ daily testing budget
Ads you can test per month (monthly budget × testing share) ÷ (multiple × target CPA)
Stop line per ad 2 × target CPA
Spend for a "3 sales" winner signal about 3 × target CPA

The logic is simple: an ad has had a fair chance once it has clearly failed or clearly started to work.

  • Clear loser: about 2× your target CPA spent without a single purchase. It would have to be far better than it has looked so far to come good.
  • Likely winner: three or more purchases at or under target. An ad selling exactly on target needs about 3× CPA of spend to get there, a better one gets there sooner.

Treat three sales as a first filter, not proof — they can be luck. That's fine, because a winner's next step is a bigger budget, where it has to keep selling to keep the money.

The testing share is a rule of thumb too. Many media buyers put somewhere between a tenth and a third of their spend into new ads: more in a young account with no proven ads, less once a few winners carry the account.

A worked example

Say you sell candles, your target cost per purchase is $5.15, and you want verdicts on 6 new ads at the default 2×. You can spend $15 a day on the test, and your monthly Meta budget is $1,546, with 20% set aside for testing.

Result Sum Answer
Test budget 6 × 2 × $5.15 $62
Days to verdicts $62 ÷ $15 4 days
Ads you can test per month ($1,546 × 20%) ÷ (2 × $5.15) 30 ads, about 7 a week
Stop line per ad 2 × $5.15 $10 spent with no sale
Winner signal 3 × $5.15 about $15 for three sales on target

Remember that Meta doesn't split an ad set's budget evenly. It pushes spend towards the ads it expects to do best, so after four days some ads will have spent far more than $10 and others almost nothing. An ad that barely spent is untested, not beaten. If you still believe in it, bring it back in a later test.

Choosing the multiple

Multiple What it buys Use it when
1× A quick, cheap first look: enough for about one sale if the ad sells on target Budgets are tight and you'll re-test anything promising
2× (default) Enough for a clear loser to show itself Most tests, most of the time
3× Room for a winner selling on target to reach three sales You want firmer winners before you scale

Many media buyers work at 2–3×. Judging earlier saves money but makes verdicts noisier: you'll stop some ads that would have sold.

How to act on the result

  • If verdicts take longer than a week, test fewer ads at once or raise the daily budget. Long tests get muddied by fatigue, sales and changing offers. Keep at least three ads in every test, so there's something to compare.
  • Spread the monthly number across the weeks. 30 ads a month is about seven a week — one test at a time. The ad variation calculator helps you decide which seven.
  • Judge on the two lines. Pause an ad at 2× target with no sale. Call a winner at three or more sales at or under target, then scale it in steps. Give unclear ads another 24–48 hours.
  • Don't wait for the learning phase to end. Meta says an ad set needs about 50 optimisation events in a 7-day window to exit the learning phase. Most tests won't get there, and they don't need to: you're comparing ads inside the test.

The stop line here is the default that BulkAds' Autopilot uses. In the tests it launches itself, Autopilot pauses an ad that spends 2× your target CPA with no sale (you can move the line anywhere from 1.5× to 3×), calls a winner after 48 hours with three or more sales at or under target, and raises its budget one step without going past the daily limit you set — asking you first, by default.

For the reasoning behind these numbers and a weekly testing rhythm, read how many ad creatives to test.

Questions people ask

How many new ads can I test each month?

Take the part of your monthly spend you set aside for new ads and divide it by what one ad needs before a verdict, usually 2–3× your target cost per purchase. The calculator does the sum and splits it into a weekly number.

How long should one ad run before I judge it?

Until it has clearly failed or clearly started to work. A common rule of thumb is about 2× your target cost per purchase with no sale for a loser, and three or more sales at or under target for a likely winner.

When should I turn off a losing ad?

A common line is once it has spent about 2× your target cost per purchase without a sale. That's also the default BulkAds' Autopilot uses, adjustable from 1.5× to 3×.

How long should a creative test run?

Until every ad has a verdict, which usually takes a few days to a week. If the calculator says it will take longer, test fewer ads at once or raise the daily budget.

Does a creative test need to exit the learning phase?

No. Meta says an ad set needs about 50 optimisation events in a 7-day window to exit learning, which most tests won't reach. You're comparing ads within the test, not waiting for the ad set to settle.

Keep going

The tool runs entirely in your browser. Results are estimates to plan with — your own account’s numbers always win.

BulkAds AI is an independent product and is not affiliated with, endorsed or sponsored by Meta Platforms, Inc. Facebook and Instagram are trademarks of Meta Platforms, Inc.

Your next fifty ads are one link away.

Paste your website, meet your brand kit and see your first batch in minutes. 7 days for $9.99 — cancel anytime.

yourbrand.comMake my ads

See pricingExplore featuresBrowse the library