The Scaling Protocol: Our Copyable Decision Tree for Raising Meta Budgets Without Tanking ROAS

A homewares brand we looked at had a good account and a bad habit. Every time a campaign hit target for a day, the founder doubled the budget that night. Performance would wobble, he'd panic, slash it back, and two days later he'd do the whole thing again. He'd been bouncing between $800 and $2,400 a day for months and never got past it.

The campaign wasn't the problem. He had no rule for when to push, how hard, and when to leave it alone. So every decision was a fresh gut call, and gut calls under pressure are how good accounts get wrecked.

What I want to give you here is the opposite of a gut call. It's the actual decision tree we run to scale Meta budgets, written out as a series of yes/no questions you can answer in about two minutes standing in your ads manager. No clever bid tricks. Just a flow that tells you exactly what to do next so you stop scaling on feelings.

Run it top to bottom every time you're thinking about touching a budget.

Question 1: is the account hitting target?

Start here, and be strict about it. Not "is this one campaign green today", but is the account as a whole sitting at or above your target on a proper window. I look at the last full day or two, on a 7-day click basis, against the new-customer cost per acquisition I'm willing to pay.

If yes, you're on the scale-up branch. Keep reading down.

If no, skip ahead to the scale-down branch further down this post. Don't scale up a campaign that isn't actually working just because one ad had a nice afternoon.

The reason I anchor on new-customer cost rather than blended ROAS is simple: Meta's job in the account is to bring you new buyers. Returning customers are what your email and SMS are for. If you scale off a blended number, you can be "hitting target" on the back of repeat purchases while your actual acquisition quietly bleeds. So the gate is new-customer cost, and it has to be green before anything else happens.

Question 2: has it been consistent for 48 to 72 hours?

This is the question almost everyone skips, and skipping it is what causes the tank.

One good day is not a signal. It's noise that happened to land in your favour. So before I scale, I want two to three straight days of performance sitting at or above target. We call that being in scaling protocol: once you've got that run behind you, you've earned the right to start raising budgets on a daily rhythm.

If you've only had one good day, the answer here is "not yet". You don't cut, you don't push. You wait another 24 hours and check again. That's the whole move. Waiting is doing.

There's one exception. If you're running a hard-dated promo, a Black Friday weekend or a three-day sale, the clock is your constraint, not the data. In that window you can skip the wait and scale into it, because the event ends whether your account is comfortable or not.

And one addition we've made over time: if something genuinely new has happened, a fresh winning ad that's taking off like a rocket and dragging your return from a 2x to a 4x, that's a real reason to move sooner. Even then I usually give it 36 to 48 hours to prove it isn't a one-day fluke. A new winner is permission to look earlier, not permission to skip the check.

Question 3: is the performance real, or is it view-through?

You've got your consistent run. Before you raise the budget, one last gate: is the result actually being driven by your ads, or is it attribution smoke?

Pop the attribution comparison and look at how many of your purchases are click-based versus view-through. My rule of thumb is that 60% or more of the purchases should be click-based, or your click-based numbers on their own should still be hitting target. If most of your "performance" is view-through, the ads aren't really doing the work, and if you scale into that you're pouring budget behind a number that won't show up in your bank account.

If it passes, you scale. If it doesn't, you wait another day and keep an eye on it rather than feeding it.

Question 4: how big is the step?

You've earned the increase. Now, how much?

The default, almost every time, is 20% on the campaign that's working. Hitting target, consistent, click-based: up 20%, done, move on. It's boring and it's right. Surfing a budget up and down multiple times a day has never given us consistent results. The steady 20% step has.

The one time I go harder is when you're well clear of target. If your goal is a 2.5x and you're sitting at a genuine, sustained 3.75x, that's 50% of headroom above target, and there are times we'll raise the budget by something closer to that, even 1.5x or double it. Do that at your own risk and only on a real, repeated overshoot, not a single lucky day.

Worth flagging: the size of the step has to bend to the size of the account, and I'll come back to that at the end. A 20% jump on $200 a day is pocket change. A 20% jump on $20,000 a day is a different conversation entirely.

That's the full scale-up branch. Green on target, green for 48 to 72 hours, green on click-based data, then 20% up. Now the other side.

The scale-down branch: when the account isn't hitting target

Here's where most damage gets done, because here's where panic lives.

First question on this branch: are you actually below break-even, or just below your ideal? There's a real difference. If your target is a 2.5x and you're running at a 2x, you're not where you want to be, but you're not losing money either. That's not an emergency, it's a slightly-off day.

If you're not below break-even, treat it like Question 2 in reverse: has it been consistently below target for 48 to 72 hours? One ugly day buys nothing but a wait. You leave it alone and check tomorrow. More often than you'd believe, it auto-corrects on its own and you'll be glad you didn't touch it. Only after a genuine three-day-ish run below target do you step the budget down, and even then only by that same 20%.

If you are properly below break-even and have been for a day or two, then yes, ease the budget down 20% to take the pressure off while you work out what's wrong. The point is that cutting is a measured 20% step on real, sustained underperformance, never a panic-halving into a single bad morning.

Notice the asymmetry built into the tree. I'll scale up off a single good day once I've earned the run. I won't scale down off a single bad one. Up and down are not the same speed.

The hard deck: the floor that makes patience safe

"Leave it alone and wait" only works if you've set yourself a floor, otherwise a genuinely cold stretch can quietly bleed you. That floor is what we call a hard deck.

A hard deck is a daily spend level you decide in advance and never drop below, whatever the numbers do. When a campaign goes cold, you don't crash it back toward zero, because at near-zero you can't test your way out and you've basically switched the account off. You step down to your hard deck and you hold there while you go fix the real problem.

The level is yours to set against your runway. A newer brand with tight cash might hold a modest floor. An established account with reserves can hold a much higher one, because it can afford to keep buying data while the creative team works. The number isn't the point. Deciding it with a clear head, in advance, is the point, so that on a bad day you're following a plan instead of reacting to a feeling.

What you actually do while you're holding the deck

Here's the part the decision tree points to but can't do for you. When an account is stalling, the answer is almost never inside the ads manager. It's outside it.

When you scale and keep getting punched, it's usually because the creative can't convert the colder, bigger audience Meta now has to reach. So sitting in the account refreshing numbers fixes nothing. While you hold your hard deck, the work is: launch new creative tests, double-check the funnel for anything broken (out-of-stock, a snapped cart, a stray bit of code), process what's actually working so your next ads are sharper, and test a genuinely new offer if you've been flat for a while. That's what gets the inside of the account moving again. The budget button never was the lever.

Adapting the tree to your spend level

The flow is the same at every size. The aggression isn't.

At a few hundred dollars a day, you can be bolder. The dollar swings are small, so a 20% step, or even a bigger one on a clear overshoot, barely registers in cash terms. Keep it simple: one campaign, lock your creative and your landing page, run the tree.

As you climb into five figures a day and beyond, the same 20% becomes thousands of dollars a move, so you get gentler. We'll often step bigger budgets up by 15%, 10%, even 5%, and lean harder on the 48-to-72-hour patience, because at that scale a clumsy jump costs real money. The questions don't change. The step size and your nerve do. Don't copy the budget moves of an account spending ten times what you are. Earn that scale, then loosen the dial.

Where to from here

If you run this tree honestly for a month, the bouncing stops. You scale when you've earned it, you hold when you haven't, and you stop turning normal dips into self-inflicted ones.

If you'd find it useful to have someone trace your account against this flow and show you where you've been pushing too early or cutting too hard, that's a fair chunk of what a Signal/Noise Audit does: it lines up your real new-customer numbers, your click-based split and your scaling history so the right next move is obvious rather than a guess. No obligation at all. Even just pulling your last few budget changes and checking each one against these questions will tell you plenty about where your protocol is leaking.