Why customer acquisition cost rises as spend on one channel grows

Average cost hides the cost of the next customer

Average acquisition cost divides all spend by all customers. A budget decision turns on the return from the next increment of spend, which is the marginal measure Meridian's documentation describes.

A hypothetical case shows the gap. A channel takes $10,000 and returns 100 customers, an average cost of $100. The budget is doubled to $20,000 and returns 160 customers. The new average is $125, which looks like a modest slide.

The extra $10,000 bought only 60 customers, so the marginal cost is about $167 each. If a customer is worth $150 in revenue, the first block returned $1.50 per dollar and the second returned about $0.90. The blended number never shows that the second block lost money.

What the response curve says about the next dollar

Google's Meridian documentation defines the pieces. A response curve shows how a channel's incremental outcome changes with its spend, with other channels held constant. Spending more, the documentation says, yields "progressively smaller returns."

Meridian separates two measures. ROI is incremental outcome divided by the channel's total cost over the historical period, and it grades past performance. Marginal ROI, or mROI, is the return on the next small increment of spend above the historical level.

The documentation reads the gap between the two as a saturation signal: when mROI is much lower than ROI, the channel is beginning to saturate at historical spend level. Channels with the highest mROI are the best for investing additional funds.

The arithmetic above follows the same logic: ROI is the average and mROI is the return on the next increment, so a large gap between them is the saturation signal the documentation describes.

The five zones between first dollar and last

A Google and Ekimetrics piece published in April 2026 describes the relationship between spend and gain as S-shaped rather than a straight line. It divides the curve into five zones.

In wear-in, revenue grows more slowly than spend because of early friction and long purchase cycles. In acceleration, growth is steepest. In efficiency, past the inflection point, marginal returns diminish but still lift average ROAS.

In contribution, average ROAS has peaked and is falling, yet each added dollar still adds revenue and profit. In saturation, the curve flattens and the next sale costs more than it brings in.

That sequence explains a common confusion. Average efficiency begins falling in the fourth zone, while the extra spend is still worth making. Falling average ROAS alone is not proof that a budget is too large.

The piece sets the profitability floor at a marginal ROAS of 1 on turnover. Above it, continued investment pays on a revenue basis; below it, the saturation point has been passed.

What Target CPA guidance says about low targets

In Target CPA bidding, the system aims to get more conversions for a target cost per action the advertiser sets. The help page warns that a target significantly below the campaign's historical average may not be attainable while maintaining reasonable levels of traffic, and says that when the budget is too low for the target, the advertiser can either raise the budget or lower the target.

The page ties a low target to traffic: a target significantly below the historical average CPA may not be attainable while maintaining reasonable traffic, so it says you should consider raising your target.

The page does not give a threshold for how far below history is too far, and it does not quantify the volume lost.

When mROI is much lower than ROI, the channel is beginning to saturate at historical spend level.

How to see it in a company's own numbers

A response curve shows the point at which a channel becomes saturated, past which further investment yields diminishing returns.

What it changes about budget decisions

Three decisions follow from the curve. The first is where to put the next dollar. By Meridian's logic, it goes to the channel with the highest marginal ROI, not the one with the best average.

The second is when to stop. A falling average is not by itself a reason, since the contribution zone still adds profit. The reason is a marginal return that no longer covers its cost.

The third is whether the problem is the budget at all. Google and Ekimetrics argue that creative quality and cross-channel synergy can shift the whole curve upward, producing more sales for the same spend.

The authors also caution against stopping at peak ROAS, which they say often leaves significant revenue on the table. Their figures come from a meta-analysis of 83 marketing mix modelling studies in European retail across Germany, Spain, France, Italy and the UK, covering July to December 2023 through January to June 2025. On average, they report, a retail brand could theoretically raise media investment by up to 329% before reaching saturation, and by 830% in search.

Those numbers carry limits. The piece is published by an advertising platform, it covers European retail only, and its footer notes that actual results will vary by advertiser. They suggest that saturation can be further away than a falling average suggests. They do not show where any one company's curve flattens, and that is a measurement question for the company's own data.