Ad Spend Budget Allocation: 70/20/10 and Marginal CAC (2026)

Ad spend budget allocation that follows marginal return

Ad spend budget allocation is how you split a fixed media budget across channels, campaigns, and tests so the next dollar buys the most profitable demand. Fixed percentage splits (“Meta 50 / Google 50”) look tidy and often waste money once one channel saturates.

A workable rule: fund each pocket until the next incremental customer approaches your CAC or contribution target, then stop and move dollars to the cheaper next customer elsewhere.

Start from the business number, not the platform

Before you open ads managers, lock three inputs:

1. Revenue or new-customer target for the period.
2. Contribution margin after COGS, payment fees, and variable fulfillment.
3. Allowable CAC / break-even ROAS (see How to Calculate ROAS and What Is ROAS in Advertising).

Goal-back math:

Required media budget ≈ needed conversions × target CPA

If that budget exceeds cash, either cut the target, improve conversion rate / AOV, or accept a longer payback—do not silently lower quality just to hit a lower CPA.

A practical split: 70 / 20 / 10

Once the total is set, allocate by role—not by habit:

• 70% proven — campaigns that clear contribution targets at current volume.
• 20% scale candidates — channels or angles with healthy early signals but not yet proven at spend.
• 10% experiments — new placements, creatives, geos, or offers with an explicit kill date.

Rebalance weekly. Money in the 70% bucket that no longer clears margin should move, even if the channel “always got half.”

Channel roles in the mix

Map spend to funnel jobs:

• Capture — high-intent search / shopping: protect coverage on money terms; do not starve this for cheap awareness.
• Create — paid social / video / discovery: buy attention, then measure assisted and blended outcomes, not only last-click ROAS.
• Convert leftovers — remarketing and CRM: usually higher ROAS, lower incremental lift; cap so it does not cannibalize the whole budget.

For Meta-style pacing inside a campaign structure, see Facebook Campaign Budget Optimization. For search-side pacing discipline, see Google Ads Budget Pacing.

How to reallocate without thrashing learning

1. Compare marginal CPA / MER, not vanity CPI alone — cheap installs or clicks that never activate are not “efficient.”
2. Respect minimum learning spend — starving a campaign below delivery thresholds creates noise that looks like “the channel failed.”
3. Move in steps — shift 10–20% of a loser toward a winner, wait for a clean read, repeat.
4. Separate prospecting vs remarketing when you judge winners — mixed budgets hide which job is failing.
5. Keep a reserve — 5–10% unallocated for seasonality, creative hits, or sudden CPC spikes.

Measurement stack for allocation decisions

Use three layers so platform ROAS does not dictate the whole pie:

1. Platform — day-to-day delivery and creative diagnostics.
2. MMP / warehouse attribution — cross-channel install and purchase truth where available.
3. Blended MER + contribution — company-level check that allocation is not cannibalizing organic or inflating refunds.

When layers disagree, shrink the disputed channel’s share until measurement is fixed—do not scale the loudest dashboard.

Weekly allocation review checklist

• Which pockets beat allowable CAC on a contribution basis?
• Where did incremental spend last week fail to buy incremental outcomes?
• Is any channel below learning volume because budget was sliced too thin?
• Did remarketing share creep above the cap?
• Which experiment hit its kill date with no path to the 20% bucket?

Write the new split in one table, update budgets the same day, and log the reason. Allocation without a written reason becomes politics.

Common ad spend budget allocation mistakes

• Copying last year’s percentages after product, geo, or creative mix changed.
• Judging every dollar on last-click ROAS while brand / social creates the search demand.
• Spreading budget across too many campaigns so none exit learning.
• Cutting the only capture channel to “save money” when demand still exists.
• Treating lower CPI as success while activation and D7 retention fall.

FAQ

Should every channel get equal budget? No. Equal splits ignore different marginal returns and learning minimums.

How often should allocation change? Weekly for digital performance accounts; sooner only for clear breakage (tracking loss, policy limits, creative collapse).

Is MMM required? Helpful at larger spend. Until then, disciplined contribution math + blended MER already beats static percentage splits.