Discounts, break-even ROAS and allowable CPA

Compare the original and discounted order

Illustrative inputs: original price $100, discounted price $80, and unchanged non-ad variable costs of $50. Original contribution before ads is $100 − $50 = $50, giving break-even ROAS of $100 ÷ $50 = 2×. Discounted contribution is $80 − $50 = $30, giving $80 ÷ $30 ≈ 2.67×.

Check the amount left after acquisition

At an illustrative CPA of $20, the original order leaves $50 − $20 = $30, while the discounted order leaves $30 − $20 = $10. To produce the same $300 contribution after ads, you would need $300 ÷ $30 = 10 original-price orders or $300 ÷ $10 = 30 discounted orders. This assumes unchanged CPA and costs; it is not a prediction that the discount will deliver that volume.

Recalculate fees and fulfillment

Some payment fees depend on revenue, while packaging and shipping may be mostly unchanged. Bundles can change weight, picking costs, and the number of units in an order. Do not assume every cost stays fixed or falls in proportion to the discount. Use the actual fee schedule and fulfillment estimate for the promoted basket.

Measure the campaign on a consistent basis

If your conversion event already sends the discounted amount, do not subtract the discount again in a reconciliation. Match refunds to the same order cohort. Compare contribution after ads, order counts, and new-customer counts alongside ROAS, because the ratio alone can conceal a change in the value of each sale.

Decide what would justify the promotion

Write down the specific objective: clearing inventory, acquiring new buyers, or increasing current-period contribution. Future purchases may justify a different budget only when supported by an explicit model and relevant data. Before launch, calculate the minimum acceptable contribution and the cost of a disappointing outcome. Afterward, compare actual orders and costs with those assumptions instead of judging the promotion from revenue alone.

Examples are illustrative, not client results or benchmarks. The calculations use the inputs shown; real-world decisions depend on your records and assumptions. Report a correction.