Do New-Customer Discounts Improve Meta Ads Profit? Build the Test Before the Offer
A new-customer discount has to earn back the margin it gives away. More orders are not enough. I would compare the contribution generated per eligible visitor, then subtract the cost of bringing those visitors in. That exposes an offer that makes conversion rate look better while leaving less money in the business.
Calculate the required lift before launching the offer. Then test whether your customers deliver it. Keep new-customer eligibility, returns and repeat purchases in the same readout so the first sales spike does not decide the whole budget.
Define contribution before comparing offers
Start with the selling price after discounts and refunds. Subtract product costs, packaging, payment fees, shipping subsidy and expected return handling losses. That is contribution before advertising, not net profit. Fixed overhead still needs covering.
For this test, define an eligible visitor consistently and assign each visitor to one offer. Count first orders from new customers under that assignment. Contribution per eligible visitor equals total contribution from those first orders divided by all eligible visitors, including people who buy nothing. It is not revenue per visitor or profit per customer.
Use the same identity and eligibility rules in both groups. Record unrecognized returning customers, coupon sharing and visitors who encounter both offers. Those are possible contamination, not evidence that every discounted order was incremental.
Calculate the lift the discount needs
All figures below are hypothetical. Assume one first order per converted visitor, the same $60 variable cost per order in both groups, and no later purchases in this initial comparison. The cost includes an expected allowance for returns. Replace these assumptions with your own product mix.
| Input or result | Full price | 20% discount |
|---|---|---|
| Net selling price per order | $100 | $80 |
| Variable cost per order | $60 | $60 |
| Contribution before ads per order | $40 | $20 |
| Eligible visitors | 10,000 | 10,000 |
| New-customer conversion rate | 3% | 4.5% |
| First orders | 300 | 450 |
| Contribution before ads | $12,000 | $9,000 |
| Contribution before ads per visitor | $1.20 | $0.90 |
Conversion improves by 50%, but contribution per visitor falls by 25%. At equal media cost per visitor, the discount loses. To match the full-price contribution, the discount needs a 6% conversion rate: $1.20 divided by $20. That is twice the original 3% rate, or a 100% relative lift.
The general calculation is full-price conversion rate multiplied by full-price contribution per order, divided by discounted contribution per order. If discounted contribution is zero or negative, more first orders cannot restore positive first-order contribution.
Add media cost and later purchases
If both groups cost $0.70 per visitor, contribution after advertising becomes $0.50 per visitor at full price and $0.20 with the discount. If advertising the offer changes traffic cost, use each group's actual cost. A cheaper visitor could change the decision, but it does not belong in the model as an assumed benefit.
Check later contribution at equal cohort ages, such as day 30, 60 and 90. Include repeat orders, returns and the costs of serving those orders. Divide by the original assigned visitors when comparing the total value of the test; contribution per acquired customer can be a separate diagnostic.
Wait for actual repeat purchases before using them to justify a larger discount. The new-customer CAC guide explains why blended ROAS cannot settle this question. Use the bundle worksheet when basket composition also changes.
Keep the experiment focused on the offer
I would agree the primary metric, budget exposure, eligibility and readout window before launch. Randomly assign eligible visitors to consistent offer experiences when the test setup supports it. Keep product availability, shipping terms and the purchase path comparable. Verify assignment and order joins before interpreting a winner.
A website offer test measures the offer among arriving visitors. A comparison of separate ad campaigns can also change who arrives and what the traffic costs. Record which question you are testing. Two campaigns run side by side are not automatically a controlled experiment.
Review sample balance, coupon leakage, email promotions and return maturity. A short run with a handful of purchases is not a firm answer. Choose the sample requirement around the smallest commercially useful difference and observed variation, then avoid repeatedly stopping the test when a favorable result appears.
Bring the next offer to the account review
I prepared creative and budgets before managing $544,397.42 in BFCM Meta spend for one US jewelry brand. An offer needs that same preparation: clear economics and decisions the team can execute.
If you spend at least $15,000 a month on Meta, review my ecommerce service and schedule a call. Bring the current offer, order costs and acquisition numbers. You work directly with me.