The short answer: never put a discount in the first abandoned cart email, reserve codes for the last message in the sequence, make every code unique and expiring, and exclude people who don’t need one — repeat customers and shoppers who abandoned over something other than price. Handled this way, a discount is a closing tool you spend on the minority of carts that need it. Handled the default way — 10% off blasted to every abandoner within an hour — it becomes a standing tax on your margin that trains customers to abandon carts on purpose.

You already suspect the discount is leaking money

Most store owners in the €20,000–100,000 monthly range set up their cart flow once, usually with a “here’s 10% off” template, and it works — orders come in, the dashboard attributes revenue to the flow, everyone moves on. Then you notice things. The same customer redeems the cart code on her third and fourth orders. Your best-selling product, which people clearly want at full price, is going out the door at 10% off over and over. Meanwhile your Meta CPMs are up again this quarter, so the margin on each order was already thinner than last year before the code shaved another slice off.

That’s the real pain: the discount isn’t recovering carts you would have lost. Much of it is subsidizing carts you would have recovered anyway.

Why “just discount more” backfires

When recovery numbers dip, the tempting fix is a bigger code — 15% instead of 10%, free shipping on top. But price is only one of several reasons people abandon. Surprise shipping costs, checkout friction, comparison shopping, and simple distraction all produce identical abandoned carts. A discount only addresses the price-sensitive slice, and it addresses everyone else too, at your expense. Worse, a predictable code creates a learned behavior: savvy shoppers add to cart, close the tab, and wait for the email. You can’t measure that directly, but every merchant who has run an always-on 10% code for a year has felt it.

What an automatic discount actually costs

Illustrative numbers, so run your own: suppose your flow recovers 60 orders a month at an average of €70, and every one carries a 10% code. That’s €420 a month in discounts — €5,040 a year. If even half of those buyers would have completed without a code (returning customers, people who abandoned at the shipping step, people who came back from the first plain reminder), you’re handing out roughly €2,500 a year for nothing. On a store running 25–30% net margin, that’s a meaningful chunk of profit spent solving a problem that mostly wasn’t price.

The rules, in the order to apply them

1. Strip the discount from your first email today. Message one should remove friction — show the cart, answer objections, link straight back to checkout. Many carts close on this alone. What that email should contain is covered in what the first abandoned cart email should say.
2. Move any code to the final message. In a three-email sequence, the code appears — if at all — in email three, 24–48 hours after abandonment, framed as a last call.
3. Make codes unique and expiring. One-time codes generated per recipient, valid for 48 hours. A static code named CART10 will be on a coupon site within a month, applied to orders that never saw your email.
4. Exclude people who don’t need it. Branch the flow: repeat customers and high-intent segments get the full sequence without a code; the code branch is for first-time visitors only, or for carts below a value threshold where margin allows it.
5. Later, not now: tiered offers by cart value and predictive discounting. Get the basics stable first; sophistication on top of a leaky default just leaks more precisely.

The flow, specified precisely

• Trigger: cart abandoned, no purchase within 1 hour. (The flow must exit immediately on purchase — nothing burns goodwill like a discount email arriving after someone paid full price.)
• Email 1 — 1 hour after abandonment, everyone: cart contents, one clear button back to checkout, answers to the two most common objections (shipping cost, returns policy). No code.
• Email 2 — 24 hours, non-purchasers: social proof or product detail — a review of the carted item, your delivery time, your guarantee. Still no code.
• Split before email 3: has this contact placed an order before? Yes → email 3 is a plain “your cart expires soon” note, no code. No → email 3 carries a unique 10% code expiring in 48 hours, subject line naming the deadline, one button.
• Channel: email throughout; SMS for the final message only on carts above roughly €100, where the discount cost is justified by the order size.
• Goal: recover the maximum number of carts while spending discount budget only on the segment that plausibly needs it.

For the decision logic behind when a discount belongs in the sequence at all, see when to add a discount to an abandoned cart flow — and if you want to test how far you can get with none, recovering carts without a discount is the companion piece.

A concrete before-and-after (invented numbers)

Picture a home goods store: 400 abandoned carts a month, average cart €70. Old setup: 10% code in email one, 55 recoveries, €385/month in discounts. New setup: no code until email three, and only for first-timers. Emails one and two now recover, say, 35 carts at full price; the final code email adds 15 more, of which only the 10 first-timers used a code — €70 in discounts. Total recoveries dropped slightly (50 vs 55), but discount spend fell from €385 to €70, and full-price recoveries rose from 0 to 40. On margin, the new flow wins clearly. Every number here is illustrative; the point is that recovered-order count alone can’t tell you which flow is better.

Metrics to watch

• Full-price recovery rate — recovered orders with no code redeemed, divided by abandoned carts. This is the number the default setup hides.
• Discount cost per recovered order — total code value redeemed divided by recovered orders.
• Margin per recovered order, not just revenue — a €70 order at 10% off and free shipping can be worth half of what the dashboard implies.
• Code redemption by customer type — if repeat buyers are redeeming cart codes, your exclusion branch is broken.

Doing this in Omnisend

Everything above needs three capabilities: conditional splits in the automation, unique single-use discount codes, and an exit condition on purchase. Omnisend covers all three — its workflow builder branches on order history and cart value, and it generates unique Shopify discount codes with expiry inside the email editor, so you’re not pasting a static code into a template. On my own stores I ran the lazy version first: static 10% code, email one, for months. I’d dismissed email as outdated spam for years, and when ad costs forced me to take it seriously I tested Klaviyo and Omnisend; I stayed with Omnisend because rebuilding the flow with splits and unique codes was something I could do myself in an afternoon. Fair warning: the tool won’t decide your discount policy for you, and if your margins can’t absorb 10%, no flow structure fixes that — fix the offer economics first.

Your next step

Export last month’s orders that redeemed your cart-recovery code and count how many came from repeat customers. That single number tells you how much the default setup is costing you — and whether the exclusion branch should be this week’s project.