Budget allocation is one of the most consequential decisions in a Meta Ads account, and one of the least systematically approached. Most advertisers either concentrate all budget on their best-performing campaign (leaving scale on the table) or spread it too thinly across too many campaigns (denying any single campaign the volume it needs to optimise). The 70-20-10 budget allocation framework solves this by creating a principled structure for how to deploy budget across three distinct strategic purposes.

The Budget Allocation Mistake Killing Most Accounts

The most common mistake: allocating 100% of budget to what is currently working, with nothing reserved for testing. The result: when the winning campaign begins to fatigue (and it will — every campaign does), there is no tested alternative to replace it. The account goes into crisis mode, testing under time pressure with insufficient budgets to generate statistically reliable data. Performance collapses while you scramble for a new winner.

The second most common mistake: testing so many things simultaneously that no single test gets enough budget or time to produce conclusive data. A budget of £50 per day split across 10 creative tests delivers meaningless data. The same budget concentrated on 2 to 3 tests with clear hypotheses produces actionable insights.

The 70-20-10 Framework Explained

70% — The Scale Bucket. Budget allocated to campaigns with a proven track record of delivering at or above your target ROAS. These campaigns have sufficient data to be trusted, proven creatives with strong performance histories, and audiences that are converting at sustainable CPAs. The goal of the scale bucket is not to find new winners — it is to extract maximum revenue from proven ones.

20% — The Test Bucket. Budget allocated to structured tests with clear hypotheses. Each test is designed to find the next entrant for the scale bucket. Tests include new creative variants (hook testing, format testing, copy angle testing), new audience strategies (new lookalikes, interest expansions, broad targeting), and new campaign structures (objective testing, bidding strategy comparison).

10% — The Explore Bucket. Budget allocated to genuinely experimental work — things that may not work at all but that could produce outsized returns if they do. New ad formats (Meta's newest features), new audience types, completely different creative directions. This bucket accepts a higher failure rate in exchange for the potential for breakthrough findings.

Budget Allocation Formulas by Monthly Spend Level

The 70-20-10 framework provides the percentage split, but the absolute budget allocation depends on total monthly spend. Here are precise allocation formulas for different spend levels:

£5,000/month accounts: £3,500 to scale proven performers, £1,000 to structured testing (typically 2 to 3 tests), £500 to exploration. At this volume, the test and explore budgets may be underutilised — consider consolidating to 70/25/5 or even 75/20/5 if you lack high-confidence tests to run.

£15,000/month accounts: £10,500 scale, £3,000 testing, £1,500 exploration. At this level, you can run 3 to 4 simultaneous tests in the test bucket and have sufficient exploration budget to test new placements or creative formats.

£30,000/month accounts: £21,000 scale, £6,000 testing, £3,000 exploration. At this level, the testing budget now allows for parallel testing on multiple fronts: 2 to 3 creative tests, 1 to 2 audience tests, and structured bidding strategy comparisons simultaneously.

£60,000+/month accounts: £42,000+ scale, £12,000+ testing, £6,000+ exploration. At this scale, the testing and exploration budgets are large enough to maintain multiple concurrent test programmes. Advanced accounts run test budgets across 4 to 6 distinct test initiatives without starving any single test of sufficient budget.

How to Allocate Within the 70 (Scaling)

Within the scale bucket, allocate by ROAS efficiency — not by absolute ROAS. A campaign with a 6× ROAS but limited audience size cannot absorb unlimited budget. A campaign with a 3.5× ROAS but a broad audience and strong scaling efficiency may be a better allocation target despite lower absolute performance.

Use the concept of marginal ROAS: as you increase budget on any campaign, at what rate does ROAS diminish? Some campaigns scale efficiently to 5× their starting budget before efficiency degrades. Others degrade quickly. The scaling allocation should prioritise campaigns with shallow marginal ROAS curves — those that maintain efficiency as budget grows.

The Dynamic Reallocation Trigger: When a Campaign Outperforms

A campaign moving from good to exceptional performance is the primary trigger for budget reallocation within the scale bucket. Define the trigger mathematically: if a campaign achieves ROAS 20% higher than its historical average for 5 consecutive days, trigger a reallocation review. Reassess: is this performance sustainable or temporary? Does the campaign have audience reach remaining, or have we saturated the responsive portion of the audience? What campaign is currently underperforming relative to its historical average and could be reduced to fund the outperformer?

The reallocation rule: increase the outperforming campaign by 15 to 20% while decreasing an underperforming campaign by the same percentage. Maintain this for 7 days and assess the results. If the outperformer continues exceeding its elevated budget while the reduced campaign stabilises or improves, the reallocation is permanent. This systematic approach captures upside without abandoning campaigns that may be temporarily underperforming due to seasonal or algorithmic fluctuations.

How to Allocate Within the 20 (Testing)

Each test within the 20% bucket should have a clear hypothesis ("If we change the hook from a question format to a social proof format, we expect CTR to improve by at least 20%"), a defined measurement period (typically 7 to 14 days), and a predetermined decision rule ("If the test variant achieves the hypothesis outcome, it moves to the scale bucket").

Avoid testing everything simultaneously. Prioritise tests in the order that will have the largest impact on overall account performance: creative tests first (highest leverage), audience tests second, campaign structure tests third, bid strategy tests last.

The 70/20/10 Testing Rule in Practice

The testing bucket is not "10 different tests at 2% each" — it is a structured testing programme with clear prioritisation. The testing rule: run 1 high-priority test at 50% of the test budget, 1 to 2 medium-priority tests at 25% each, and reserve 0% for opportunistic exploration until the high-priority test produces results. High-priority tests are those with the largest potential impact on overall account ROAS if successful. For most accounts, creative testing (especially hook and copy angle testing) is consistently high-priority because creative changes affect all campaigns.

Case Study: Budget Allocation Transformation and ROAS Impact

Client: UK e-commerce brand, £20,000/month spend

Before reallocation (month 1): Account was running with 90% of budget on two top campaigns, 10% scattered across testing. Blended ROAS: 2.8×. The account had tested nothing in 60 days and was entirely dependent on the two proven campaigns. Both campaigns showed signs of fatigue (declining marginal ROAS on weekly budget increases). Testing budget was too small to run any meaningful tests.

Implementation (month 2): Shifted to 70/20/10. Scale bucket (£14,000): maintained the two proven campaigns at slightly reduced budgets, allocated the recovered £6,000 to three other previously under-tested campaign structures. Test bucket (£4,000): ran 2 creative tests (£2,000 each) — one testing new hook angles, one testing image formats. Exploration bucket (£2,000): tested Reels placements exclusively.

Results (month 3): The creative hook test produced a winner — a specific problem-articulation hook outperformed the baseline by 31% on CTR. This winner was promoted to the scale bucket (reducing one fatiguing campaign). Total blended ROAS improved to 3.4×. Cost per acquisition decreased from £185 to £165. The Reels exploration test identified a placement that, while lower volume, converted at 1.8× the cost of the baseline placements — allocated more budget based on the finding.

Month 4-6 compounding: With a structured testing programme generating winners, the account maintained the improved ROAS while total monthly spend was scaled to £25,000 (previously growth was limited by lack of tested alternatives). The consistent testing discipline transformed the account from dependent on two campaigns to distributing scale across five distinct creative and audience combinations — significantly reducing single-campaign risk while improving overall efficiency.

Budget Pacing: Daily vs Lifetime Settings

Daily budgets give you consistent visibility into daily spend and allow easy incremental adjustments. Lifetime budgets give Meta's algorithm more flexibility to concentrate spend on high-converting time windows (evening hours, weekends for consumer products) but are harder to monitor and adjust.

Our recommendation: daily budgets for scale campaigns where predictability and control matter most. Lifetime budgets for time-sensitive campaigns (product launches, seasonal promotions) where Meta should have maximum flexibility to concentrate spend when conversion probability is highest.

Scaling Triggers and Increment Sizes

Scale campaigns should have predefined triggers — the conditions under which you increase budget — to remove emotion from scaling decisions. Our standard trigger: if a campaign maintains target ROAS for 5 consecutive days at the current budget level, increase budget by 20 to 25%. Wait 3 days before the next increase to allow the algorithm to restabilise. Never increase budget by more than 30% in a single step — larger increases can destabilise algorithmic learning and cause temporary performance drops.

Rebalancing When the Market Shifts

The 70-20-10 allocation is not a set-and-forget structure. Market conditions change: a new competitor enters, CPMs spike due to seasonal demand, iOS privacy changes reduce targeting precision. When performance in the scale bucket degrades across multiple campaigns simultaneously (not just one), it is a signal to rebalance — temporarily reduce the scale bucket allocation and increase the test bucket to find new approaches adapted to the new conditions.

Quarterly rebalancing reviews are standard practice for well-managed accounts. At each review: assess the health of the scale bucket, promote test-bucket winners, retire underperforming scale campaigns, and ensure the explore bucket is producing learnings rather than just spending budget. This quarterly rhythm prevents the allocation from becoming stale and ensures the account is always progressing toward higher efficiency.

The Marginal ROAS Concept and What It Means for Scaling

Marginal ROAS — the ROAS generated by each additional pound of budget added to a campaign — is the most important concept in budget scaling decisions, and one that almost no advertiser calculates. A campaign running at £2,000 per month with a 5× ROAS does not necessarily maintain a 5× ROAS at £4,000 per month. As budget increases, the campaign reaches deeper into the audience, often finding less pre-qualified prospects who require more creative exposure before converting. The marginal ROAS on the additional £2,000 might be 3×, bringing the combined ROAS down to 4×.

Calculating marginal ROAS: increase a campaign's budget by 20% for one week. Calculate the ROAS on the incremental spend specifically. If your £2,000 campaign at 5× ROAS generated £10,000 in revenue, and after increasing to £2,400 it generates £11,400, the marginal ROAS on the additional £400 is £1,400 ÷ £400 = 3.5×. This is your marginal ROAS. If 3.5× is above your break-even threshold, the increase is profitable and further scaling is warranted. If it is below, you have found the campaign's efficient budget ceiling at this stage.

The Testing Budget: Why Most Advertisers Get It Wrong

Most advertisers treat testing budget as waste — money spent on things that may not work. The reframe: testing budget is the investment required to manufacture the next winner. Without testing, the scale budget is operating on the inevitable decline of whatever is currently working. With a healthy testing programme generating winners at a consistent rate, the scale budget always has fresh high-performers to allocate to. The 20% testing allocation is not an expense — it is the R&D investment that keeps the 70% scale allocation productive.

The testing budget mistake: running 10 simultaneous tests with £50 per day each. With £5 per day per test, most tests never reach the statistical significance needed to make a confident decision. The result: inconclusive data, no clear winner, and budget consumed without generating learnings. The correct approach: run 2 to 3 focused tests simultaneously with £20 to £30 per day each. Concentrated budget generates conclusive data faster and produces actual winners more efficiently than spreading the same budget across 10 thin tests.

Seasonal Budget Rebalancing

The 70-20-10 framework is a year-round baseline, but specific periods demand rebalancing. Q4 (October to December) for consumer e-commerce brands typically warrants shifting to 80 to 85% scale and reducing testing to 10 to 15% — the demand environment is so strong that aggressive scaling of proven performers generates higher returns than testing during the period. January to March typically warrants the opposite: increased testing allocation as demand normalises and you prepare the new winners needed for the year ahead.

Service businesses tend to have different seasonality: B2B decision-making picks up in January and September (new budget cycles, new planning periods), making these the ideal months to run new campaign and audience tests. The findings from these high-intent testing windows provide the winners to scale during the mid-year and end-of-year revenue pushes. Aligning your testing and scaling allocations to your market's natural demand rhythm amplifies the effectiveness of both.

Communicating Budget Strategy to Stakeholders

For in-house teams and agency-managed accounts, the 70-20-10 framework provides a clear communication structure for budget decisions. Rather than explaining individual campaign decisions, you can frame every allocation decision in terms of the strategy: "This £500 increase goes into the testing bucket to identify the next performer for scaling." "This £2,000 reallocation moves budget from a fatiguing scale campaign to a tested winner." The framework turns individual budget decisions into evidence-based implementations of a coherent strategy — which builds stakeholder confidence and reduces the reactive "why are we spending on X?" conversations that waste everyone's time.

Advanced Budget Allocation: Account Maturity Model

The 70-20-10 split assumes a mid-stage account. Adjust based on account maturity: Early-stage accounts (under £2,000/month): Use 60/30/10 split. Early-stage accounts lack proven campaigns, so testing allocation is higher to find winners faster. Growth-stage accounts (£2,000 to £10,000/month): Use 70/20/10 split. This is the equilibrium point where you have winners to scale and sufficient budget to test new angles. Mature accounts (£10,000+/month): Use 75/15/10 or 80/10/10 split. Mature accounts have deeply optimised winners with strong scaling efficiency. Testing budget can be lower because new variables have diminishing returns.

Budget Reallocation in Response to Ad Fatigue

Ad fatigue is the inevitable decline in campaign performance over time as audiences see the same creative repeatedly. Recognise fatigue signals: CTR decline of 15%+ over 2 weeks without budget changes, CPA increase while audience size remains constant, viewthrough rate declining. When fatigue is detected: rotate creative immediately (new hooks, angles, formats), reduce frequency (fewer impressions per user per day), expand audience (broader targeting, new lookalikes). Do not reallocate budget away from a fatiguing campaign immediately — reallocate only after new creative is tested and proven superior.