A founder we work with disagreed with their media agency about testing budgets. The founder set the budget per creative. Each new ad got a fixed amount, enough to buy a few conversions at the target CPA. The total followed from the number of creatives, whatever the month’s overall spend. The agency set testing as a share of total spend. A bigger month meant a bigger testing budget.
Both sides had a rule. Both sides lacked a test result that showed which ads worked.
Both camps are half right
Operators on the Marketing Operators podcast use both rules at once.
Rolain, at HexClad, sets a floor per test: “We try to spend at least 10x AOV on the product we’re pushing as a baseline.” A creative that gets less money has not been tested.
Herrmann, of Herrmann Digital, sets a cap on the portfolio. He puts 10% of overall spend into testing. Unproven creative never takes more than 20% of the budget at any time. The other 80% goes to known winners.
The two rules answer different questions. The floor sets the price of one readable test. The cap sets how many tests you can afford. The founder’s rule covers the first question. The agency’s rule covers the second. The two rules together give the number of creatives you can test each month, from your spend and your AOV.
A test below the floor tells you nothing
HexClad found this out in its own account. The team used to launch 7 to 10 ad sets a week. With a high AOV, each ad set got only 8 to 10 conversions in a seven-day window. Meta’s algorithm needs more purchase volume per ad set to learn. The team cut to 3 to 4 ad sets a week and moved to fewer, more distinct concepts. It still launches hundreds of variations a month.
Statistics point the same way. Lewis and Rao studied 25 advertising field experiments. Individual sales vary so much that most tests could not reject a zero effect. Only three of the 25 could tell a 50% ROI from zero. Small tests produce noise that looks like a result.
Meta’s own delivery has changed too. Operators report that Advantage+ and the Andromeda update spread spend more evenly across good creatives in a campaign. One or two winners used to take 80% to 85% of a campaign’s budget. Brands now need a deeper bench of good-enough ads.
More ads do not mean more winners
Some operators answer with volume. Saamir Mithwani, a dropshipper, claims 30 creatives a day across his portfolio. Dante Timpano, another dropshipping operator, reports the other side of that bet. At 150 or more ads a week, his hit rate was 0.8% to 1%. It rose to 5% to 6% after he cut volume and put more intent into each ad. His winner is an ad that spends over $10,000 at target KPI for at least two weeks. He reports these figures himself on a podcast, without audited data. His definition is still worth copying. He sets the spend and the time a winner needs before the test starts.
Two dropshippers also report cheap kill rules for new products. Andrew Yu claims he kills a product after about $100 to $150 of loss. Mithwani claims a break-even product deserves 30 to 40 creatives and up to $150 to $200 of cumulative loss. Those caps help probe a product. Seb de Lophem sets a much higher bar for a brand. He calls a brand validated once it can spend $2,000 to $3,000 a day profitably, consistently.
Most of a testing budget should not go to brand-new ideas either. Operators cite an 80/20 rule from Tube Science. Teams put 80% of creative output into iterations on winning concepts and 20% into new ones. HexClad runs closer to 50/50.
Work out your testing budget this week
You need your AOV, your monthly Meta spend and the number of new creatives your team ships each month.
- Set the floor. Multiply your AOV by 10. That is the minimum spend behind each new creative before you judge it.
- Set the testing budget. Take 10% of monthly Meta spend. Check that unproven creative never passes 20% of spend in any week.
- Divide the testing budget by the floor. The result is the number of creatives you can test properly this month.
- Compare it with what your team produces. You judge some ads on noise when your team ships more creatives than you can test. Cut production, or spend the surplus on iterations of proven winners. You can afford more new ideas when your team ships fewer.
- Write the winner rule before launch: spend level, KPI and duration. Track your hit rate each month. Add volume only while the hit rate stays flat or improves.
- Log every test, including the ads you cut, with the reason.
A brand with a low AOV gets a low floor and can test many creatives. A brand with a high AOV and a modest budget can afford only a handful each month.
From a winning ad to its own page
A creative that passes the floor and hits your KPI has proven an angle. Most brands then send that angle to the same product page as every other ad. On a binding-lever call, we look at your winning angles and the pages they send people to. We tell you which one deserves its own page first, or whether your next step is more creative. Book the call.