The Hidden Costs Your ROAS Doesn't See

A campaign spends $100,000 and generates $500,000 in attributed revenue. The dashboard reports a 5.0× ROAS. Marketing scales it. Then finance closes the month: returns arrive, payment processors take their fee, and shipping, fulfillment, packaging, labor, platform, and vendor costs sit outside the advertising dashboard.

The campaign still generated $500,000 in revenue. It just did not generate nearly as much profit as the ROAS implied. That is not a problem with ROAS. It is a problem with asking ROAS to answer a question it was never designed to answer.

ROAS measures the efficiency of buying revenue

ROAS = attributed revenue ÷ ad spend. Spend $100,000 and generate $500,000 in attributed revenue, and ROAS is 5.0×. It is useful for comparing campaigns, creatives, audiences, and channels. But every dollar of revenue is treated as equal: full-price and heavily discounted orders, high- and low-margin products, retained and later-returned orders.

ROAS sees the purchase. It does not see the full economics of fulfilling it.

The distance between revenue and true margin

Consider the same campaign after its orders move through the business.

The 5.0× ROAS is still mathematically correct. But the business generated $145,000 of contribution before advertising and retained $45,000 after paying for the campaign. The costs are not literally hidden; they are hidden from the metric being used to make the acquisition decision.

Replace revenue with contribution: POAS

For the campaign above, POAS = $145,000 ÷ $100,000 = 1.45×. ROAS says the campaign generated $5.00 in revenue per dollar spent; POAS says it generated $1.45 in contribution before advertising, leaving $0.45 after paying for ads.

A POAS of 1.0× is the first-order break-even point when its numerator is contribution before advertising. The definition must be explicit: usually net revenue less COGS, payment processing, pick-and-pack, fulfillment, shipping subsidies, packaging, returns processing, marketplace/platform/order-level app fees, variable warehouse labor, vendor fees, and marketing accruals. Vendor allowances, funding, shipping revenue, and restocking fees may offset these costs.

The exact list changes with the operating model. There is no universal true-margin field waiting inside a commerce platform; the business has to define it.

Do not hide arbitrary fixed-cost allocations inside campaign performance

Facilities, salaried labor, software contracts, and other fixed costs determine company profitability, but they do not always belong inside POAS. A warehouse lease does not change because one campaign generated an additional order. Allocating it across campaigns can help company-level planning but can distort the incremental acquisition decision.

• POAS uses consistently defined direct and variable costs to evaluate acquisition.
• True margin can include allocated operating costs to show what remains at the company, channel, product, or customer level.

The first order is not the customer

POAS improves the first-purchase decision, but repeat customers require another lens. One channel can have lower first-order ROAS but acquire customers who return, buy at full price, purchase across categories, and rarely return merchandise.

Contribution LTV:CAC = lifetime contribution margin before acquisition cost ÷ customer acquisition cost. The acquisition cost stays out of the LTV numerator because it is already the denominator.

Paid social wins on the metric visible inside the ad platform. Organic search creates the more valuable customer.

Four metrics, four decisions

ROAS: can we buy revenue efficiently?

Use it for tactical campaign monitoring and platform bidding, especially when product margins and return rates are similar.

POAS: was the first order worth buying?

Use it when margins vary across products, returns are material, shipping is subsidized, or fees meaningfully change order economics.

Contribution LTV:CAC: was the customer worth acquiring?

Use it to compare acquisition channels, customer segments, first products, discount strategies, and cohorts over a defined horizon.

True margin: is the business model profitable after operating it?

Use it to understand the full margin waterfall, including the operating costs required to run the company at its current scale. Optimizing only ROAS favors the campaign that buys the most revenue; the full ladder favors customers, products, and channels that leave money behind.

Building the measurement layer

Advertising platforms alone cannot calculate this reliably. The required data spans orders, discounts, returns and exchanges; product costs and vendor terms; payments and processor fees; shipping, fulfillment and warehouse activity; marketing spend and attribution; customer identity and repeat purchases; and platform, marketplace, and app fees.

Sources need to join at the correct grain. Order-level costs attach to orders, product costs to units sold, returns back to the original order and campaign, and customer activity across purchases without double-counting identities. Then every layer needs a written definition: net revenue, exchange treatment, shipping charges, variable warehouse costs, return timing, and LTV horizon.

Without those decisions, POAS becomes another precise metric that different teams calculate differently.

Frequently asked questions