Facebook Ads Budget Allocation: The 70/20/10 Rule

Ask ten media buyers how they split their Facebook ad budget and you'll get ten answers. But there's one rule that keeps surfacing in serious accounts because it works at almost every stage of growth: the 70/20/10 rule.

It's borrowed from Google's old innovation framework, but it maps surprisingly well onto paid media. Once you start using it, you'll stop having those existential moments where you're frantically rebuilding the entire account because two ad sets stopped working.

What the rule says

70% of budget goes to proven, profitable campaigns
20% of budget goes to scaling — pushing winners harder, expanding to new audiences and markets
10% of budget goes to true experimentation — new concepts, new formats, new angles

For a brand spending £1,000/day, that's £700 on what's working, £200 on growth bets, and £100 on weird experiments that might or might not pay off.

It's a discipline more than a formula. The point is to stop flying by the seat of your pants and start thinking about portfolio risk.

Why this matters more than it sounds

The biggest reason Facebook ad accounts collapse is over-concentration. You find one killer ad. You scale it. You scale it more. You forget about creative refresh. You forget about new audiences. Two months later the ad fatigues, performance tanks, and you've got nothing in the pipeline because you never built anything new.

The 70/20/10 split forces you to keep planting seeds. The 10% experimentation budget is where next quarter's winners come from. Without it, you're not running paid media — you're harvesting until the field goes barren.

What goes in the 70%

The 70% is your stable, profitable, boring core. Things that have been running for at least 21 days at a positive ROAS. Proven creatives, proven audiences, proven offers.

Don't touch this aggressively. Increment budget by 15-20% every 5-7 days if performance allows. Refresh creative inside ad sets rather than rebuilding ad sets. The rule for the 70% is: if it ain't broke, don't fix it.

Specific things in the 70%:

Your top 2-3 winning ads at the established budget
Your retargeting layer (BOFU) at proven spend
Your DPA campaign if you're ecommerce and it's working

What goes in the 20%

The 20% is for scaling and expansion. This is where you take a winner and ask, "how do I get more of this?"

Duplicating winning ads into new lookalike audiences
Expanding a winning campaign into a new country or region
Testing higher budget tiers on proven creatives
Launching CBO versions of currently ABO-running ad sets

You're taking calculated risks here, but with proven assets. The downside is limited because the underlying creative has already proven it can convert.

If the 20% scaling tactics work, they get promoted into the 70% stable bucket and the cycle repeats.

What goes in the 10%

This is where the magic happens — and where most advertisers cheat themselves by skipping it.

The 10% is for genuinely experimental stuff:

New creative concepts you have no proof will work
Untested angles or hooks
New formats (Reels, Stories, Carousels you've never tried)
New audience types (lookalikes built off different seeds)
Whole new offers or product positioning

Most of the 10% will fail. That's the point. You're not betting on individual experiments — you're betting on the portfolio of experiments. One in five will produce a winner. That winner gets promoted to the 20% bucket, and eventually to the 70% bucket.

If you don't run a 10% experimentation budget, you're effectively living off your last winning creative until it dies. That's not a strategy, that's a slow-motion catastrophe.

How to apply the rule at different account sizes

The rule scales but the absolute numbers change behaviour.

£50/day total: £35 stable, £10 scaling, £5 experimentation. The £5 is too small for serious testing, so let it accumulate over a week and run a £35 test once a week instead.

£200/day total: £140 stable, £40 scaling, £20 experimentation. Now you can run a 3-2-2 test regularly with the experimentation bucket.

£1,000/day total: £700 stable, £200 scaling, £100 experimentation. Healthy. You can test 5-8 new concepts per month and have meaningful data on each.

£5,000/day total: £3,500 stable, £1,000 scaling, £500 experimentation. You should be running constant testing cycles and have a creative team feeding the pipeline.

When to violate the rule

There are two situations where 70/20/10 doesn't apply.

Brand new account. You have no "proven" anything yet. For the first 4-6 weeks, you might be running 30/10/60 — most of your spend is testing because you're trying to find product-market fit on the platform.

Major creative refresh moment. If your entire winning creative bank has fatigued at once (it happens, especially seasonally), you might temporarily push 40% into the experimentation bucket to find new winners faster.

Outside these two situations, 70/20/10 should be your default split.

How to enforce the split in practice

Most media buyers don't enforce the split because it requires discipline that's easier in theory than reality. A few practical tactics:

Label every campaign in your naming convention with [Stable], [Scaling], or [Test]
Once a week, calculate actual budget split across the labels and rebalance if you're outside the rule
Cap your test budget so you can't accidentally over-spend on experiments and starve the stable layer
Run a monthly review of what got promoted from Test → Scaling → Stable

It's a small operational habit that compounds massively.

Outsourcing the discipline

If you'd rather not manually track which campaigns belong in which bucket, Pix-Vu does the rebalancing automatically. It tags campaigns by performance stage, allocates spend across the 70/20/10 split, and promotes winners between tiers without you needing to do the bookkeeping. £99/month — cheaper than the budget you'll waste forgetting to do this manually.