Set a target CPA from contribution margin
Calculate an acquisition ceiling from contribution after discounts, product cost, fulfillment, fees and a chosen reserve, with a worked USD waterfall.

Start with the dollars available from one completed order
For a first-order acquisition decision, calculate how much contribution remains before advertising, then decide how much of it the business needs to retain. The difference is an operating ceiling for acquisition cost. It is a planning boundary derived from your inputs, not a universal target CPA.
This method uses one completed new-customer order as its unit. Revenue is merchandise sales after discount and expected merchandise refunds; tax collected for remittance is outside this example. Product, fulfillment, shipping subsidy, variable service and payment costs are included. Fixed overhead and future repeat purchases require separate treatment.
A worked contribution waterfall
Consider a fictional product with a $120 listed price and a $12 discount. All amounts below are illustrative USD. The $4 refund allowance is an explicit planning assumption per acquired order, including its chosen effect on variable cost; it is not an observed return-rate benchmark.
| Step | Amount | Contribution remaining |
|---|---|---|
| Listed merchandise price | 120 | 120 |
| Discount | −12 | 108 |
| Expected net refund effect | −4 | 104 |
| Product cost | −38 | 66 |
| Fulfillment and shipping subsidy | −12 | 54 |
| Payment and variable service cost | −6 | 48 |
| Desired contribution after advertising | −18 | 30 available for acquisition |
Before advertising, the order contributes $48. Retaining $18 leaves a maximum planned acquisition cost of $30: $104 − $38 − $12 − $6 − $18 = $30. The $18 is a management choice to support overhead and other needs. It is not a claim that the example business earns an $18 net profit.
If you want a revenue-based cross-check, $48 ÷ $104 is a pre-ad contribution rate of approximately 46.15%. Spending the entire $48 would consume all of that contribution. Retaining $18 restricts spend to $30, equivalent to $104 ÷ $30 = approximately 3.47× revenue-to-ad-spend under this exact revenue policy.
Make sure CPA counts the action that pays the costs
A cost per form submission cannot be compared directly with a ceiling per new paying customer. If only one in four accepted inquiries becomes a first-time buyer, the customer-level ceiling has to be translated using that explicitly measured funnel. Keep incomplete cohorts and duplicate inquiries out of the conversion estimate.
For an illustrative 25% accepted-lead-to-new-customer rate, the $30 customer ceiling corresponds to $7.50 per accepted lead before any additional variable sales expense. If handling a lead costs $2, that expense also needs a place in the model. A cheaper lead is not automatically an affordable customer.
Google describes target CPA bidding as aiming for an average cost per conversion; individual costs vary. That platform setting is distinct from the financial ceiling calculated here, and its conversion definition must be checked. [Google Ads: About Target CPA bidding]
Test the costs that change with the order
A promotion can change several rows at once. A $10 deeper discount reduces eligible revenue and the available acquisition budget if other inputs remain unchanged. A heavier product can increase fulfillment cost. A larger basket can improve contribution dollars even when a percentage margin looks similar. Rebuild the waterfall for the actual product mix.
- Use realized net revenue for a mature cohort or state a provisional refund allowance. Do not subtract both for the same adjustment.
- Separate fixed subscription software from a transaction fee that increases with every order.
- Record whether shipping charged to the buyer is included in revenue and whether delivery expense is included in variable cost.
- Check whether free samples, commissions or sales handling add an acquisition-related variable cost.
If contribution before advertising is zero or negative, the first order supplies no positive acquisition allowance under this model. A future-value strategy may still be considered, but it needs evidence about repeat contribution, timing and cash. Adding an optimistic lifetime value estimate does not repair a missing first-order cost.
Carry the policy into the campaign review
I would write the revenue policy, included variable costs and retained-contribution requirement beside the ceiling. Then compare it with the same customer-level acquisition measure from a mature cohort. When the achieved cost exceeds the ceiling, investigate price, cost, mix and conversion before assuming the bidding control alone can solve it.
Sources
AI-assisted writing and an AI-generated editorial cover. Source links and checked dates are listed below; illustrative examples are labeled in the text.
- Google Ads: About Target CPA bidding — checked 2026-10-18