Found Money: What Reallocating Your Testing Budget Does to Annual Revenue (With a Calculator)

The advice you've heard a hundred times is "ring-fence 20% of your budget for testing." The reality nobody mentions is that most of that 20% is being set on fire on purpose, and you're calling it strategy.

I want to put an actual dollar figure on that fire. Not a vibe, not a "you're probably wasting some" hand-wave. A number you could write on a whiteboard.

So let's build the calculator together. I'll use a made-up brand that spends a tidy A$1,500 a day, which is A$45k a month, because round numbers make the maths easy to follow. Plug your own figures in as we go.

The way most accounts are set up

Here's the standard structure, and odds are it's yours.

You've got a testing campaign where you force new creatives to spend. Roughly 20% of your daily budget goes here. On A$1,500 a day, that's A$300 going into testing.

The other A$1,200 a day runs through your "proper" campaigns, the ones with your proven winners in them.

Now the bit people skip over. Your testing campaign almost always returns a worse ROAS than the rest of the account. It's supposed to, in theory, because it's full of unproven creative. Say your testing sits at a 1.5x return and your main campaigns sit at a 2.0x. Those are gentle, believable numbers, not fantasy.

Let's run it.

• Testing: A$300/day at 1.5x = A$450/day in revenue
• Everything else: A$1,200/day at 2.0x = A$2,400/day in revenue
• Total: A$2,850/day, which is roughly A$85,500 a month on A$45k of spend

That's a 1.9x blended ROAS. Fine. Lots of brands would take it. But hold that A$85,500 in your head, because we're about to find money that was sitting in plain sight.

Why forcing budget into testing is the leak

The problem isn't that you test. You have to test. The problem is how the budget gets to the new creative.

When you force A$300 a day into a testing campaign, you're guaranteeing that money spends, whether the creative deserves it or not. A dud gets the same A$300 push as a future winner on day one. You're paying full freight to learn something the algorithm could've told you for a fraction of the cost.

Here's my take after watching this play out across a lot of accounts: you don't need to force-feed testing budget anymore. If you put new creative into the same environment as your winners and let it earn its spend, the platform only pours money into the new stuff once it's proven it can compete. The losers quietly starve. The winners get fed.

The practical difference is that your genuinely bad creative stops eating a fixed 20% of your budget. Instead it eats whatever scrap it can earn before the algorithm gives up on it, which in a well-fed campaign is closer to 5% of your spend. Sometimes less.

That gap, 20% down to 5%, is the leak. Let's price it.

Recomputing with the budget where it should be

Same A$1,500 a day. Same creative. Nothing about your product, your offer, or your team has changed. We've only moved where the money sits.

Now your weakest creative pulls about 5% of budget instead of 20%. That's A$75 a day, not A$300. And I'll be harsh on it: let's say that A$75 returns the same poor 1.5x it did before. No improvement assumed.

• Bottom creative: A$75/day at 1.5x = A$112.50/day
• Your proven winners now run with the freed-up budget. That's A$1,425 a day. The top slice of that behaves like your good campaigns always did, around 2.0x.
• A$1,425/day at 2.0x = A$2,850/day
• Total: roughly A$2,962/day, or about A$88,875 a month

So we went from A$85,500 to A$88,875. Call it A$3,375 a month in extra revenue, for changing nothing except where the dollars flow.

I want to be careful here, because this is exactly the spot where marketing maths gets dishonest. That A$3,375 is not free profit, and it's not a promise. It's the modelled upside of redirecting wasted spend toward creative that's already shown it works, holding every ROAS assumption flat. Your real numbers will wobble. But the direction is the point, and the direction is reliable.

Now compound it

A month of found money is a nice-to-have. A year of it is a different conversation.

That A$3,375 a month is A$40,500 over twelve months. On its own that's already a meaningful number for most 6-7 figure brands. It might be a hire. It might be your entire creative production budget. It might be the difference between a stressful Q1 and a calm one.

But it gets better, and this is where founders perk up. Because that extra revenue isn't disappearing into thin air. If your unit economics let you put it straight back into ads, you're now spending A$48k a month instead of A$45k, with the same structure and the same efficiency. Which throws off a bit more, which you reinvest again.

I won't pretend that compounding is infinite, because efficiency does soften as you scale, and I'd be lying if I drew you a hockey stick. But even the flat, un-compounded version, A$40,500 a year from a structural change you make once, is the kind of thing that's been hiding in your account the whole time. You just couldn't see it, because it was disguised as a sensible 20% testing rule.

How to actually move the money

Reading the maths is easy. Doing it without nuking your account takes a bit of care. Here's roughly the sequence I'd follow.

Don't dump new creative into a winning ad set. This is the one that catches people. If you drop fresh ads into an ad set that already has learnings, you reset those learnings and tank what was working. New creative goes into its own new ad set, every single time.

Let new ads compete in the same campaign as your winners, not a separate ring-fenced one. The whole trick is that the algorithm distributes budget to whatever's most likely to convert. Give it your new stuff and your proven stuff in one place, and it'll back the new creative only when it earns it. That's the mechanism that turns your 20% leak into a 5% trickle.

Expect a slower read on new creative, and make peace with it. Forced testing tells you in 3 to 5 days, mostly by burning cash. The earn-its-spend approach can take 7 to 14 days, because the creative has to climb the ladder. Slower, yes. But you're not paying a flat A$300 a day for the privilege of finding out.

Graduate winners deliberately. When a new ad proves itself, promote it into your scaling campaign on purpose. Don't leave it sitting in the testing pool forever. The winners are what your spend should concentrate on.

None of this requires new software or a bigger budget. It's the same dollars, routed with more discipline.

Run your own numbers before you trust mine

Everything above used a clean A$45k-a-month brand because it made the arithmetic readable. Your account won't be that tidy. Your testing might be at 18% or 25%. Your ROAS spread might be wider or thinner. The leak might be smaller than my example, or, honestly, a fair bit bigger.

So do the five-minute version yourself. Take your monthly spend. Work out what share is genuinely going to unproven creative. Multiply that share by the gap between your testing ROAS and your proven ROAS. That difference, annualised, is your found money. Write it on the whiteboard and sit with it.

If you run that exercise and the number makes you wince, that's usually a sign the leak has been quietly compounding against you for a while, not for you. The good news is it cuts both ways once you fix the structure.

And if you'd rather not eyeball it alone, that's part of what a Signal/Noise Audit is for. We map exactly where your spend is going, what's earning its keep, and what's draining budget under the banner of "testing", then put a real annual figure on the gap. No pressure either way. But at minimum, go do the whiteboard maths today. The number tends to be larger than people expect, and it's been sitting there the whole time.