Cross-Channel Budget Pacing: When to Shift Spend Between Google and Meta

The best-performing account of the month is rarely the one with the best creative. It’s the one whose budget was in the right place at the right hour. Pacing is the skill nobody puts on their resume — and the one that quietly decides whether you hit target or explain a miss.

Average ROAS is the wrong number to allocate on

Open any benchmark report and Google looks like the obvious place to put money. Across more than 16,000 US search campaigns measured from April 2024 through March 2025, Google Ads posted a median 7.52% conversion rate at a $5.26 average CPC, according to WordStream’s 2025 benchmarks. Meta’s lead campaigns landed close on conversion rate — 7.72% — but at a far cheaper $1.92 CPC and a $27.66 cost per lead versus Google’s $70.11, per WordStream’s 2025 Facebook data. Two different pictures, and neither tells you where the next dollar should go.

Averages describe the past. They pool every dollar you already spent — the cheap early conversions and the expensive late ones — into a single blended figure. That number is fine for a board slide and useless for a budget decision, because the decision is always about the next dollar, not the last thousand. To pace across two platforms, you need to stop asking “which channel performed better?” and start asking “which channel will convert my next dollar better?”

What marginal ROAS actually measures

Marginal ROAS is the revenue produced by one additional unit of spend. It isolates the impact of the next dollar and, critically, exposes diminishing returns that a blended average hides — as Lifesight defines it, it answers how effective your next dollar will be rather than how effective your spending has been. The formula is simple: divide the change in revenue by the change in spend.

Here is why it flips decisions. Take an operator running both platforms and deciding where an extra $2,000/day belongs. Google carries the higher average ROAS by a wide margin — but watch what the next increment actually returns.

Illustrative figures.Google “wins” on average ROAS 4.2x to 2.3x. Yet its next $2,000 returns only 2.5x while Meta’s returns 3.2x. The incremental spend belongs on Meta — the platform that looks worse on the report. Allocate on the average and you pour money into a saturating channel; allocate on the margin and you buy the cheaper conversions that are still available.

Reading the diminishing-returns curve

Spend does not convert to revenue in a straight line. Every channel moves through three phases — increasing returns, linear returns, then diminishing returns — as measurement firm Measured describes it. Saturation isn’t a malfunction; it’s a fundamental feature of every channel. The only variable is where your account sits on the curve today.

The signal is unmistakable once you look for it. Ruler Analytics walks through a prospecting campaign where lifting spend from £55,500 to £60,200 grew revenue — but the marginal ROAS on that new spend was 5.22, roughly half the 11.05 overall ROAS. The campaign was still profitable. It had also crossed into the part of the curve where each new pound bought noticeably less than the last. That gap between the average and the margin is the tell.

When your marginal ROAS drops to half your average ROAS, the channel is telling you it’s full. Feed the next dollar somewhere with room to grow.

This is the practical read for pacing. When a platform’s marginal return sinks well below its own average, further spend there is buying expensive, low-quality incrementality. That is your cue to move money to the channel that still has slope left — not because the saturated one has “stopped working,” but because your budget can work harder elsewhere.

The equimarginal rule: where the next dollar goes

There is a clean rule for when to stop shifting. Budget is optimally allocated when the marginal ROAS of the next dollar is equal across every channel — the equimarginal principle. As long as Meta’s next dollar returns more than Google’s, keep moving spend toward Meta; the act of doing so pushes Meta down its curve and eases Google back up its own. You stop when the two marginal returns converge. That convergence point is where total return peaks.

The upside of getting this right is not marginal itself. Measured’s case study realigned a $50M budget toward higher-marginal channels and produced a 13.4% incremental revenue gain — about $16.3M — with zero additional spend. Same money, better placement. Reallocation is the cheapest growth lever you have, because it costs nothing but attention.

One caveat keeps this honest: you can only equalize marginal ROAS if you can estimate it, and platform-reported revenue tends to flatter the platform reporting it. Meta and Google each count conversions on their own model and attribution window, so their in-app ROAS numbers are not directly comparable. Anchor the comparison to a shared outcome — blended ROAS from GA4 or your CRM, or cost per qualified lead — and estimate the margin from a real spend change you made, not from the dashboard’s default view. The framework is only as good as the number you feed it.

This is the same logic behind running the two platforms as a single system rather than two silos — a theme we cover in running Google Ads and Meta Ads together. The budgets are connected whether or not your tooling treats them that way.

Pacing without underspending or blowing the budget

Knowing where spend belongs is half the job. The other half is landing the month on target — not 15% under because you were cautious, not 20% over because a good day ran away from you. This is where platform mechanics matter.

On Google, an average daily budget is a monthly instrument in disguise. Google can spend up to 2x your average daily budget on any single day, but will not bill more than 30.4x that budget across a calendar month. So a $100/day budget is really a ~$3,040/month budget that Google is free to spend unevenly. And as of June 2026, Google proactively paces toward that full monthly amount, even for campaigns on limited ad schedules — meaning a channel that ran quiet mid-month can spend hard later to catch up. Plan for the shape of the month, not a flat daily line.

A pacing check you can run in five minutes

1. Compute month-to-date pace. Divide spend-to-date by the fraction of the month elapsed. Running at 40% spend on day 20 of 30 means you are pacing to underspend by a third.
2. Check the margin on each channel.Estimate marginal ROAS from your last spend change. Whichever platform’s next dollar returns more is where the catch-up spend goes.
3. Adjust in small steps.Nudge budgets, don’t overhaul them — large jumps distort both pacing math and platform learning (next section).
4. Guard the ceiling. Watch that no single high-demand day quietly consumes the buffer you need for the rest of the month.

Underspending is not the “safe” error, either. Every dollar you leave unspent below your profitable marginal-ROAS threshold is revenue you chose not to buy. Pacing is a two-sided discipline: don’t blow the budget, and don’t leave profitable spend on the table.

How often to rebalance

The instinct is a big Monday-morning reshuffle. That instinct is expensive. On Meta, budget changes larger than roughly 20% can reset an ad set’s learning phase, forcing it to re-accumulate the ~50 optimization events per week it needs to exit learning. A dramatic weekly swing can hand you days of volatile, unoptimized delivery — the opposite of what a pacing correction is supposed to buy.

The resolution is frequency and restraint together: read the numbers constantly, move in small increments constantly. Small, frequent adjustments keep every channel near its optimal point on the curve without tripping learning resets. This maps to the same discipline behind automated bidding strategies and cutting wasted spend — steady correction beats heroic intervention.

Big weekly swings feel decisive and cost you learning. Small hourly nudges feel invisible and compound.

This is precisely the work that does not scale by hand. Estimating marginal ROAS on both platforms, checking month-to-date pace, and moving budget in sub-20% increments across dozens of campaigns — every hour, across every account — is a full-time job no operator can actually do manually. Adriva runs it as a background process, rebalancing budget between Google and Meta every hour on marginal return, which we break down in how Adriva rebalances budget hourly. The framework in this piece is the logic; the value is running it continuously instead of once a week.

The bottom line

Cross-channel pacing comes down to three moves. Judge each channel by its marginal return, not its average. Shift spend toward the higher margin until the two platforms converge. And do it in small, frequent steps so you land the month on target without resetting what the algorithms have learned. The account that wins the month isn’t running better ads — it’s running the same ads with its budget in the right place, every hour.

• Marginal beats average for every allocation decision.
• Equalize the margin across Google and Meta to maximize total return.
• Move small and often to pace cleanly and protect learning.