The 10% Rule: How We Decide Which Ads Earn a Scale Budget

Your highest-ROAS ad is usually the worst one to scale.

I know how that sounds. You open the ads manager, you spot the creative sitting at a 7 ROAS, and every instinct says pour budget in. That's the one. Send it.

In reality, that ad is often a trap, and chasing it is one of the most common ways I see good accounts stall. So let me lay out how we actually decide which ads earn a scale budget - the thing we call the 10% rule - because once you see the logic, the high-ROAS reflex stops looking clever and starts looking expensive.

The number that matters is spend, not ROAS

Here's the uncomfortable bit. On Meta, spend is performance.

If an ad has spent $80 and pulled a 7 ROAS, that's not a winner yet. It's a maybe. The algorithm has had a tiny taste of it and hasn't been able to scale it. And that matters more than people think, because Meta will happily throw money at anything it believes can hold up. If that ad could keep returning a 7 at volume, it would already be spending a fortune. The fact that it isn't is the algorithm quietly telling you something you can't see from the outside: this one doesn't scale.

I call those ROAS-supportive ads. They're great. Let them run. They prop up your blended numbers nicely. But forcing budget into them is fighting the algorithm, and you'll lose that fight almost every time.

The ads worth scaling are the ones that are spending the most AND returning above your account average. High spend, high ROAS. Both, together. Not one or the other.

Why the low-spend high-ROAS ad is such a tempting trap

Picture three ads. One spent $8 at a 14 ROAS. One spent $20 at a 9. One spent $700 at a 3.2 when your account average is 2.8.

Nearly everyone points at the first two. They look like the heroes. The $700 ad looks boring by comparison.

But the $8 ad's 14 ROAS is built on what, two or three purchases? It's noise dressed up as signal. There's no statistical weight behind it. Try to scale it and the ROAS collapses the moment real budget hits, because there was never a real result there to begin with - just a lucky handful of conversions.

The $700 ad is the actual winner. It's proven it can absorb spend and still beat your average. That's a creative you've poured hundreds of real dollars into validating, and it's held. That is the asset you build on.

So the discipline is almost the reverse of the instinct: sort your ads by amount spent first, then look at ROAS. Never the other way around. The second you let ROAS lead, you start cherry-picking ads that were never going to carry weight, and you graduate ghosts into your scale campaign.

How the 10% rule actually runs

The mechanics are simpler than people expect, which is the whole point.

We keep two campaigns doing two jobs. A prospecting CBO campaign is where everything new gets tested - fresh creative goes in as its own ad set, climbs the ladder, and either earns spend or doesn't. Then a separate Advantage+ campaign is the scale environment, and it only ever contains proven winners. That's the campaign that eventually does most of the heavy lifting in the account.

Every couple of weeks, we graduate the top 10% of ads from prospecting into the scale campaign. That's it. Top spending, above-average ROAS, the genuine standouts. Sometimes that's four ads. Sometimes, if it's been a quiet fortnight, it's nothing at all - and that's completely fine. You don't owe the scale campaign a graduation just because the calendar says so. Forcing a mediocre ad up the ladder to fill a quota is how you dilute the one campaign you most need to keep clean.

I'd block a single hour for it. Open the account, sort by spend, take the clear winners up, cut the dead weight at the bottom, close the laptop. The power isn't in any one session being clever. It's in doing the same boring hour every two weeks, for months, so the account compounds.

One thing we never do: pause the winning ad in the prospecting campaign after it graduates. If it's spending well and returning well where it is, you leave it running. You never switch off something that's working just to tidy up your structure. Let it keep doing its job in both places.

The duplicate that protects your social proof

Here's the detail that trips people up, and it's worth getting right.

When a winner graduates, you don't rebuild it from scratch in the scale campaign. You hard-duplicate it across. The reason is social proof. That ad has spent weeks farming likes, comments, shares, saves - and that engagement is part of why it converts. Build a fresh copy and you reset all of it to zero. A brand-new ad with no comments has to earn that trust all over again, and you've thrown away the very thing that made the creative work.

The duplicate carries the existing engagement with it. Same post, same social proof, new campaign, new budget environment. That's how a winner keeps its edge when it moves into scale rather than stumbling at the handover.

This is also why the scale campaign is so hard to break. It's full of ads that have already proven themselves AND already carry their social proof. You're not hoping a creative works. You're moving creatives that have demonstrated, with real spend behind them, that they do.

What this looks like over a few months

The compounding is the part worth sitting with.

Early on, most of your budget lives in prospecting, because that's where the discovery happens. Over time, that flips. As more validated winners graduate and stack up in the scale campaign, more of your budget naturally flows there - to the ads that have earned it. New creative still has to climb the ladder in prospecting and fight for spend against everything already running, so nothing gets a free ride. Only the genuinely better stuff breaks through.

And because you're never force-feeding budget into unproven creative, you waste far less on the duds. Your weakest ads end up taking a tiny slice of spend instead of burning through a fat testing budget every week. That saved spend is real money you get to redeploy into more testing or more scale.

The whole system is built to take the emotion out of it. No falling for a shiny 14 ROAS on $8. No gut calls about which ad "feels" like a winner. Just spend, then ROAS, then graduate the top 10%, every two weeks, and trust the algorithm to do what it's far better at than you are: deciding where the next dollar goes.

Try it on your own account this fortnight

If you want to pressure-test this, you don't need anything from me. Open your ads manager, set your attribution to 7-day click, and sort your ads by amount spent - not ROAS. Look at what's actually carrying your account versus what's just flattering your averages on a handful of conversions. I'd bet a few of the ads you've been mentally crowning as winners have barely spent, and a couple of the unglamorous high-spenders are the real engine.

That single re-sort tends to change how founders see their whole account. Run it, see what surfaces, and if it shifts your view of which creatives deserve the budget, that's the 10% rule already earning its keep. This is exactly the discipline we run for the brands we manage, fortnight after fortnight - the results come from the repetition, not the cleverness.