Your Best ROAS Channel May Be the Wrong Place for the Next Dollar

A channel returning 5x ROAS can deserve less additional budget than one returning 3x. The decision turns on what happens to the next dollar, while the dashboard usually shows what happened to every dollar you've already spent.

That distinction matters whenever your head of growth asks where another $10,000 should go. Sorting the channel report from highest ROAS to lowest gives you a tidy answer, but it can send the money toward an audience you've already reached efficiently and can only expand into at a much higher cost.

Consider a hypothetical independent running-shoe brand selling through Shopify. All figures below are invented teaching inputs, not Shopify benchmarks or results from an actual merchant. We'll hold product mix, prices, and measurement windows constant so we can isolate the budget decision.

The average hides the expensive edge

The brand spends $20,000 on paid search and records $100,000 in attributed revenue. Its average ROAS is $100,000 divided by $20,000, or 5x. Paid social spends $20,000 and records $60,000, producing 3x.

A straight ranking makes search the obvious winner. But suppose the team has credible evidence about the next spending increment: another $10,000 in search is expected to generate only $15,000 in additional revenue, while another $10,000 in social is expected to generate $30,000.

The marginal ROAS is 1.5x for search and 3x for social. Search's historical average remains higher because its original budget captured the cheapest opportunities. Adding budget can require broader queries, higher bids, or less qualified traffic; the new opportunities don't inherit the economics of the old ones.

This example doesn't establish that social always scales better. It shows why the answer requires an estimate of the next increment rather than a ranking of historical averages.

Put contribution behind the revenue

Assume the brand keeps 50% of net revenue after product cost, fulfillment, payment fees, and expected returns, before advertising. Treat that contribution rate as a simplifying assumption; a real forecast should reflect the products and customers each channel brings in.

The additional search budget produces $15,000 multiplied by 50%, or $7,500 in contribution before advertising. Subtract the $10,000 media cost and the expected contribution change is negative $2,500.

The additional social budget produces $30,000 multiplied by 50%, or $15,000 before advertising. After the same $10,000 media cost, the expected contribution change is positive $5,000.

Under these assumptions, the break-even marginal ROAS is 1 divided by 50%, or 2x. That threshold covers the variable costs we've included and the added advertising; it doesn't guarantee that the business covers fixed overhead or meets its profit target.

But even this arithmetic can flatter the plan if you substitute platform-attributed revenue for genuinely additional revenue. A customer claimed by an ad platform may have purchased anyway, and two platforms may claim the same order. Keep those uncertainties visible instead of treating the extra decimal place as evidence.

Estimate a response, then test the decision

You rarely know marginal returns in advance. Build a range from comparable spending periods, controlled budget changes, and experiments where practical. Historical observations help generate a forecast, although changes in seasonality, price, creative, or competitors can explain apparent budget effects.

For the hypothetical social expansion, use three planning scenarios: $18,000, $30,000, and $36,000 in additional revenue. With the same contribution assumption, those scenarios yield negative $1,000, positive $5,000, and positive $8,000 after media cost. Those are scenarios, not a statistical confidence interval.

The downside now has a price. A team might accept that risk for a reversible test, especially if the test helps answer a larger allocation question. A business short on cash may require stronger evidence before committing even when the central forecast looks attractive.

Define the test before spending: the budget increment, eligible markets or audiences, outcome window, and the contribution measure you'll use. Record promotional changes and inventory constraints so that a stockout doesn't become a false diagnosis of channel saturation.

Keep the decision small enough to learn

You don't need to rebuild the whole media plan to use marginal thinking. Begin with the next discretionary increment and keep the existing budget separate from the expansion question.

• Ask finance for contribution assumptions that reflect expected returns and product mix.

• Ask channel owners for a forecast of additional revenue at the proposed budget, with a downside scenario and a clear explanation of the evidence.

• Set a review date after enough purchases have matured, and specify which evidence would justify another increase.

Moving the entire budget from search to social would be a much larger claim than this example supports. The low-cost search opportunities may still be valuable, while social's economics may deteriorate after its own next increment.

This week, add a column called expected contribution from the next $5,000 to your budget sheet. Put the assumptions beside it. That forces the allocation meeting to address the money you're about to spend, which is the part you can still change.