The 20% Rule: How Fast You Can Scale Facebook Ads

There's a number every media buyer hears in their first month and almost nobody questions: 20%. Raise a winning campaign's budget by no more than 20% per week, the advice goes, or you'll kick it back into the Facebook ads learning phase and watch ROAS collapse.

That's the 20 percent budget rule for Facebook, and like most rules of thumb in this job, it's mostly right and badly under-explained. It hides where the number comes from, the spend bands where it's the wrong number, and the failure mode it actually prevents. So when to increase ad budget on a winning campaign isn't one answer - it's a band.

What the 20% number is actually doing

The 20 percent budget rule on Facebook is a heuristic for one specific problem: every budget change re-opens the Facebook ads learning phase, and a big change resets it from scratch. That phase needs roughly 50 conversions in 7 days to clear. Re-set it and you get the same volatile, expensive first week, only with more spend on the line.

20% is the band the industry has converged on as "small enough that Meta absorbs it without flinching". Inside that step, the optimisation signal carries over. Outside it, the algorithm treats your campaign as new.

Read the curve, not the rule. Up to 20%, the risk of a hard reset is real but contained. Past 50%, you're effectively asking Meta to restart the campaign while charging you for the privilege.

Where the 20% rule is too cautious

For small accounts, 20% can be a brake on growth that doesn't need braking. A campaign spending $50/day in absolute dollars only adds $10/day at +20%. The learning-phase risk at that volume is lower than the cost of waiting another week.

Under roughly $200/day at the campaign level, we'll often run +30% to +50% steps once the winner is genuinely stable - because the absolute dollar change is small enough that Meta's signal noise doesn't spike, and the opportunity cost of moving slowly is higher than the algorithmic cost of moving fast.

This is the part the rule never says out loud: scale step-size should bend to spend, not the other way around. BAVS calls it Banded Scaling - the 20 percent budget rule with the step-size bent to the spend band. A flat "20% per week" applied to a $50/day campaign is leaving compounding on the table.

Where 20% is the wrong direction entirely

Past a certain spend ceiling - the point where your CPMs start climbing because you've saturated the buyer pool a single campaign can reach - bigger budget steps don't break learning. They break the math.

You're not getting Meta confused at that point. You're paying more to reach the same people, and the campaign quietly stops earning its keep no matter how cleanly you respect the rule. The fix is not a smaller step. The fix is a different axis.

The 20% rule answers how much budget. Vertical vs horizontal scaling on Facebook ads is the choice the rule never makes for you.

Vertical scaling is what the 20 percent budget rule governs - lift the budget on the same campaign. Horizontal scaling is widening the structure: new ad sets, new angles, new creatives, sometimes new campaigns running the proven concept into fresh territory. Past the ceiling, every dollar belongs on the horizontal axis. Vertical vs horizontal scaling on Facebook ads is the move the rule never makes for you - which is why so many accounts get stuck spending more for less.

The other number nobody mentions: cadence

The 20% rule talks about step-size. It says nothing about how often. Stack three 20% steps inside a week and you're effectively running +73%, and the algorithm doesn't care that you broke it into three "small" moves.

The cadence we run is one step, then a settle window long enough to read the result - usually a clean 7 days of stable spend before the next move. So when to increase ad budget has a two-part answer: only off a stable trend (never a single best day) and only after the last bump has cleared its own learning ripple.

The "duplicate and double" trick - copy a winner into a fresh campaign at 2x budget - is sometimes a clean way around the limit. It works because the new campaign is, to Meta, a new campaign. You're not scaling. You're testing whether the same concept holds at higher spend, on its own learning phase.

Where this falls apart

The 20% rule assumes the algorithm is the bottleneck. Sometimes it isn't.

If your creative is fatiguing, no scale rate saves you - the next budget bump just buys more impressions of an ad people have already learned to skip. If your funnel breaks at the landing page, scaling spend amplifies a leak. The 20% rule protects the algorithm; it doesn't protect a campaign whose actual constraint is upstream. This is the creative-first lens we apply on every account: the rule assumes a winning ad. Confirm that first, every time, or the rule is solving the wrong problem.

What to do with this on Monday

Treat the 20 percent budget rule as a default, not a law. The default catches you when you're tempted to double a winner on its best day. The discipline is to bend the default to the account in front of you.

1. Match step-size to spend band. Sub-$200/day, you can usually run +30-50%. Mid-spend, hold the 20%. Past the efficient ceiling, switch axes entirely.
2. Hold the cadence. One step, one settle window, then read profit (not platform ROAS) before the next move - the MER and CAC reading is what tells you whether the last bump actually worked.
3. Catch fatigue early. Scaling magnifies whatever the creative is doing. Account drift gets ugly fast at higher spend, which is the unglamorous job BAVai does at 7am every morning - so the bump you made Monday gets pulled Tuesday if it sours overnight.
4. Move to the horizontal axis on time. When CPMs climb faster than results, the answer is not a smaller step. It's a wider structure - the play we lay out in how to scale Facebook ads without losing ROAS.

If your scaling plan is +20% a week, forever, the question to ask isn't whether the rule is right. It's whether the rule still fits the account it's running on.