How Many New Ad Creatives Should You Test Each Week?

By the Monastic Media teamUpdated 1 Nov 20267 min read

The short answer: your budget sets the number

Most pages on this topic give one figure, like 5 or 10 or 20 new ads a week, sorted by spend tiers in dollars. Those figures come from large US accounts and don't carry over to a D2C brand in India selling at a different price point with a different cost per purchase.

A better question is this: how many new ads can your budget actually judge each week? An ad that never gets enough spend to show whether it sells hasn't been tested. It has only been launched. Ten new ads a week on a budget that can properly read four gives you ten guesses and no answers.

So start from your own numbers. Volume matters, but only the volume your account can read.

Work out your number in three steps

Use your last 30 days of data from Ads Manager, or from the evening KPI report if you get one.

• Step 1: Set a weekly testing budget. This is a fixed slice of spend kept apart from your proven ads. Early brands often keep it small and protect the ads that already sell. Brands that are scaling put more into testing, because fatigue eats winners faster at higher spend.
• Step 2: Set the spend each new ad needs before you judge it. A workable rule is two to three times your target cost per purchase. Below that, one lucky or unlucky sale swings the result.
• Step 3: Divide. Weekly testing budget divided by spend per ad gives the number of new ads you can read properly. That is your weekly number. Round down, not up.
• Check it against production. If your team or vendor can only deliver half that number at decent quality, the real limit is production, not budget. Fix production before you raise the target.

Typical weekly ranges by brand stage

At the early stage, fewer ads with more spend behind each one beats a wide spread. You are looking for the first angle that sells, and thin spend across many ads hides it.

What counts as a new creative

Founders and agencies often count differently, and that is how a brand ends up paying for 30 ads a month that are really three ideas. Before you agree a number with anyone, agree what each unit is.

• New concept: a different reason to buy. A new problem, a new buyer, a new proof point or a new format such as a founder story against a demo. This is where big wins come from.
• Variation: the same concept with a new hook, first frame, opening line, creator or length. This is where you stretch a winner.
• Not new: the same video with a changed colour on the caption, a different end card or a new aspect ratio. Useful, but don't count it as a test.
• Meta now groups ads that look alike when deciding delivery, so near-identical ads compete for the same people. Real differences in angle and opening matter more than raw count.

A weekly testing calendar that works

A fixed weekly rhythm keeps production and testing in step. One that suits most D2C accounts:

• Monday: launch the week's new ads into your testing campaign. Launch them together, not spread across the week, so they get a fair start.
• Tuesday and Wednesday: leave them alone. Edits restart learning, as covered in our guide to the learning phase. Watch hook rate and click-through only to spot broken ads.
• Thursday: first read. Pause clear losers that have spent their full test amount with no purchases and weak click-through.
• Friday: brief next week's ads. Build variations of anything showing early purchase signal, and write new concepts to replace the losers.
• Following Monday: move ads with steady cost per purchase into your scaling campaign and launch the next batch.
• Production has to run a week ahead of testing. If scripts get written on Monday for ads going live that same Monday, the cadence will break by week three.

Signs you are testing too many or too few

• Too many: most new ads spend almost nothing because Meta picks one early and starves the rest. Your losers were never judged, so you can't learn from them.
• Too many: quality drops. Rushed scripts, weak hooks and recycled footage that just adds noise.
• Too few: frequency on your best ads keeps climbing while click-through falls, and you have nothing ready to replace them. See what to do when frequency is too high.
• Too few: cost per purchase creeps up month on month with no single cause. That is usually slow creative fatigue across the account.
• About right: most weeks give you at least one ad worth a variation, and you rarely scramble for a replacement.

India-specific things that change the number

• Festive and sale periods: in the run-up to Diwali, wedding season and big marketplace sale events, auctions get crowded and ads tire faster. Build extra ads in the weeks before, not during.
• Language: a Hindi or regional language version of a winning concept is a real new test, because it reaches buyers the English ad never convinced.
• COD-heavy categories: a cheap purchase on Meta can still be a costly return. Judge new ads on delivered orders where you can, as covered in our guide on high COD return rates.
• Low price points: a low cost per purchase means each ad needs less spend to judge, so you can afford more new ads a week from the same budget.
• Premium products: fewer, better ads. Lifestyle ads videos and careful demos take longer to make, and each one needs more spend before you can judge it.

Where these tests should live

How you set up the account decides whether your weekly number gets a fair read. Keep testing apart from scaling, so new ads don't have to beat proven winners for spend on day one. Our Meta Ads creative testing framework covers how to run and judge each test, and how to structure a Meta Ads account covers the campaign setup. This guide covers the other half: how many ads to feed that system each week, and how to keep the supply steady.

How Monastic Media helps

One in-house team in Surat scripts, shoots and edits the ads videos and runs the Meta and Google campaigns they go into, so the people who see the numbers make the next ad. See our ads video creation

Read next

Questions