How Much of Your Discount Revenue Is Actually Wasted?

You sent out a 15% off code to your email list last month. 2,400 customers used it. Revenue looked good. The campaign was a success.

But was it?

Here's the question nobody asks: how many of those 2,400 customers would have bought anyway, at full price, without the code?

The measurement gap in Shopify promotions

Shopify gives you redemption counts. Klaviyo gives you attributed revenue. Neither tells you whether the discount was necessary for the purchase to happen.

Research from OnePoll found that 83% of consumers who use a discount code would have made the purchase without it. That means for every 100 customers who use your code, roughly 83 were already going to buy. You gave them a discount they didn't need.

For a merchant running $100K in promotional revenue, that's potentially $83K in margin given away to customers who were already committed to purchasing.

The math that changes how you think about discounts

Consider a typical Shopify promotion:

• Discount code: SPRING20 (20% off)
• Redemptions: 1,000 orders
• Average order value: $85
• Total discount given: $17,000

Traditional measurement says this campaign generated $85,000 in revenue. It looks profitable.

But if 83% of those customers would have bought anyway, only 170 orders were genuinely extra. Sales that wouldn't have happened without the promotion. The other 830 orders just cost you $14,110 in unnecessary discounts.

Your Cost Per Incremental Order, what each genuinely extra sale actually cost you, isn't $17. It's $100 ($17,000 / 170 extra orders).

If your customer acquisition cost from paid ads is $25, that "successful" promotion was 4x more expensive than just running ads.

Returns make it worse

Here's what most analytics tools miss entirely: return rates on discounted orders are typically higher than on full-price purchases. In fashion, the gap can be 25% vs. 14%.

When you recalculate after accounting for returns, many promotions that looked marginally profitable flip to unprofitable. A promotion with 170 extra orders and a 25% return rate actually generated 128 net extra orders. Your real cost per extra order is $133.

Not all promotions are bad

This isn't an argument against discounting. Some promotions genuinely earn their keep.

A welcome code that converts browsers into first-time buyers at $3.20 per extra order, when your CAC is $18, is a fantastic deal. A win-back campaign that reactivates dormant customers at a fraction of the cost of acquiring new ones is worth running.

The problem isn't promotions. It's not knowing which ones are working and which ones are eroding margin. Without measurement, you're treating every promotion the same, and you're almost certainly overspending on some and underspending on others.

What you can do about it

Start with your historical data. Pull your last 6-12 months of orders and calculate what percentage involved a discount code. Look at the return rates on discounted vs. full-price orders. Most merchants are surprised by both numbers.

A/B test your discount codes. The only way to measure real incrementality is to run a controlled experiment: show some eligible customers the promotion and others a control experience, then compare purchase rates.

Measure in dollars, not percentages. Calculate your Cost Per Incremental Order for each promotion and compare it to your CAC. A promotion with a cost per extra order below your CAC is genuinely earning its keep.

Account for refunds. Whatever your numbers say before returns, recalculate after. The answer will be different, sometimes dramatically.

The bottom line

Discount codes aren't inherently good or bad. But running them without measuring the actual extra sales they generate means you're making decisions based on incomplete data.

Most merchants have never seen their Cost Per Incremental Order. Most have never compared return rates on discounted vs. full-price orders. Most don't know what percentage of their promo spend goes to customers who would have bought anyway.

That's the gap. And closing it starts with asking one question: "Would this customer have bought without the discount?"