Budget reallocation: moving spend before returns diminish

Budget reallocation means moving ad spend from campaigns where the next unit returns little profit to campaigns where it returns more. The key measure is marginal true-profit ROAS: the profit earned by the last increment of spend, not the average. Campaigns hit diminishing returns at different points, so budgets should move weekly, in small steps, based on evidence.

Why does average ROAS mislead budget decisions?

Average ROAS blends the cheap, high-intent conversions a campaign won first with the expensive, marginal ones it won last. A campaign averaging a healthy return can be earning almost nothing on its most recent spend. Scale it further and the average falls slowly, masking the fact that every extra unit is close to break-even or below it.

That is the second form of chasing illusions. The first is ignoring hidden costs, as covered in our true-profit ROAS guide. The second is trusting averages when decisions happen at the margin.

How do you spot diminishing returns?

Look at how results changed when spend changed. You need a few weeks of data at different spend levels for each campaign.

• Cost per result rising faster than spend: doubling spend while cost per result rises sharply is the classic sign.
• Frequency climbing in the core audience: the campaign is reaching the same people again.
• Lower-value orders: average order value or margin falling as the campaign reaches less committed buyers.
• Rising returns: marginal buyers often return more, which only shows up in true-profit numbers.

A simple weekly reallocation method

1. For each campaign, estimate marginal true-profit ROAS from the last two or three spend changes.
2. Rank campaigns by marginal, not average, return.
3. Move a modest share of budget, say a tenth, from the lowest marginal campaign to the highest.
4. Check that the receiving campaign has creative capacity, scored if new.
5. Hold for a week, observe, repeat.

Small, frequent moves beat large quarterly swings. Algorithms on both Meta and Google need time to settle after budget changes, and big jumps reset learning in ways that make the data harder to read.

Definitions

• Marginal ROAS: the return generated by the most recent increment of spend.
• Saturation point: the spend level beyond which marginal true-profit ROAS falls below your floor.
• Profitable headroom: the additional spend a campaign can absorb before reaching its saturation point.

What about cross-platform reallocation?

Moving budget between Meta and Google is where attribution overlap bites. If both platforms claim the same orders, each looks better than it is, and shifting money between them can appear to improve both. Use blended MER and true-profit numbers reconciled against real orders as the referee; more in our piece on double counting. For the strategic layer above channels, SOMIN's strategy tools help teams decide which audiences and messages deserve investment in the first place.

Is it ever right to keep funding a saturated campaign?

Sometimes. A campaign at saturation may be bringing in first-time customers with strong repeat value, or building awareness in a segment you plan to grow. Those are legitimate reasons, but they should be explicit. Write down why a campaign is being funded past its profitable point and when you will review it. Without that note, saturation becomes habit.

Understanding which segments are worth that patience is a cultural and psychological question as much as a numerical one. Mindfuse, a sister consultancy focused on consumer psychology in Singapore and Southeast Asia, works on exactly the segment and tension questions that justify long-term investment.

Agencies running multi-client budgets face this at scale. The Neo360 story on somin.ai is a useful read on how agency teams use audience intelligence to guide where money goes.

A worked example with two campaigns

Take two illustrative campaigns, each spending the same weekly budget. Campaign A reports the higher platform ROAS on average. Over the past month, though, its spend rose in steps and each step bought noticeably fewer kept orders than the last; its marginal true-profit ROAS is now close to 1.0. Campaign B has a lower average but has barely changed in efficiency as spend grew, and its marginal true-profit ROAS is still well above the floor. Ranked by average, A looks like the one to grow. Ranked at the margin, B is. Moving a tenth of A's budget to B costs A very little profit and gives B room to earn more. Do that for a few weeks, watching both curves, and you converge on a split where the next unit earns roughly the same in each. That balance point is the goal.

Two cautions. First, check that B has the creative capacity to absorb more spend; a campaign running one tired ad will saturate quickly. Second, recompute the curves each week, because seasonality and competition move them.

The average tells you where you have been. The margin tells you where the next dollar should go.

How FMedia recommends reallocation

FMedia estimates a response curve for each campaign from its spend history, applies true-profit adjustments for returns, fees and discounts, and calculates marginal true-profit ROAS at current spend. Each week it suggests a small set of budget moves, with the reasoning shown: the curve, the expected profit change, and any creative capacity concerns. You approve the moves in your ad accounts. FMedia keeps score, so next week's suggestion learns from what actually happened.

The effect is less dramatic than a big reallocation project and more useful: a steady drift of budget towards where it earns profit, with fewer surprises at month end.

Frequently asked questions

What is marginal ROAS?

The return generated by the most recent increment of spend. It falls as a campaign saturates, often well before average ROAS shows any trouble.

How much budget should I move at once?

Small steps, around a tenth of a campaign's budget per week, keep algorithms stable and make the effect of each move readable.

Does FMedia move budgets automatically?

No. FMedia recommends moves with the reasoning shown. You approve and apply changes in Meta and Google.