Shopify Payment and Gateway Fees

Payment fees are the most predictable cost in ecommerce and among the most frequently misstated. The reason is that everyone knows the headline rate, and the headline rate is one of six components.

They are also deducted rather than billed. The money never arrives, so there is no invoice to prompt anyone to record it, and a profit calculation assembled from things people paid will miss it entirely.

The anatomy of a payment fee

Percentage of order value. The quoted rate, and usually the largest component.

Fixed amount per transaction. Small, constant, and the reason blended rates mislead.

Cross-border uplift. Applied when the card was issued in a different country to your account. It follows card issuance, not shipping address, so a domestic delivery paid on a foreign card still attracts it.

Currency conversion. Where the customer pays in one currency and you settle in another, there is a spread. It is rarely itemised as a fee — it appears as a slightly smaller payout.

Tax on the fee. In many markets the processing fee itself is taxable. A small percentage of a small percentage, deducted invisibly.

Dispute fees. Charged per chargeback, regardless of outcome. Winning the dispute recovers the revenue and not the fee.

Six components, one quoted number. The gap between them is the difference between a 2.9% assumption and a 4.1% reality.

Why the fixed component decides small-order economics

The per-transaction fee is the part most likely to change a decision, because its impact is entirely a function of order value.

Thirty cents on a $120 order is 0.25% — noise. The same thirty cents on a $15 order is 2%, roughly doubling what taking that payment costs. A store selling low-priced items pays a blended rate far above the quoted percentage, and the blended figure gets worse every time average order value falls.

This has direct consequences for merchandising. A single-item order at $15 and a three-item order at $45 pay the same fixed fee, so bundling and minimum-order incentives recover real money rather than just increasing revenue. It also means that any analysis using one blended fee percentage across all orders will overstate margin on small orders and understate it on large ones — precisely inverting the ranking on the orders where margin is tightest.

Alternative payment methods

Cards are not the only cost structure, and the others differ in ways that matter.

Wallets typically price close to cards but can route differently depending on the funding source behind them.

Buy now, pay later carries a substantially higher merchant fee in exchange for conversion and larger baskets. Whether that trade works is answerable: compare contribution margin on BNPL orders against card orders of similar value.

Cash on delivery has no processing percentage and a much worse profile overall — a per-order collection fee, a materially higher failure rate, and return freight on every failed attempt. COD orders need costing separately or they quietly drag down whichever segment contains them.

Bank transfers and account-to-account are cheap per transaction and shift cost into reconciliation labour instead.

The mistake is applying one fee assumption across all methods. Payment mix shifts over time, and a store whose COD share has grown ten points has a materially different cost base than the one its spreadsheet describes.

What a refund does and does not return

When an order is refunded, the revenue reverses. The fee frequently does not.

Practice varies: some gateways return the percentage and keep the fixed component, some keep everything, and it can differ by market within the same provider. There is no way to know without checking your own payout records for a refunded transaction.

At a low return rate this is a rounding error. At fifteen or twenty percent returns on a large catalogue it becomes a standing cost that no spreadsheet built from order data will ever show, because the fee retention appears only in the payout detail.

Getting it into the profit calculation

Three rules make fees behave correctly.

Above the contribution margin line. Fees scale per order, so they belong with shipping and packaging. Below the line they inflate contribution margin and therefore inflate the acquisition cost you believe you can afford.

Per order, with the real formula. Percentage plus fixed, plus the uplifts that apply to that specific transaction. Not a monthly blended rate applied evenly, which misprices both ends of the order value distribution.

With the fee tax included where your market charges it. It is the single most commonly omitted line, and it is pure margin loss.

The check that catches most errors: total the fees your calculation predicts for a month and compare against what the payout statements actually deducted. If the two agree, the formula is right. If the calculation is consistently lower, one of the six components is missing.

When it is worth renegotiating

Payment rates are negotiable above a certain volume, and the threshold is lower than most stores assume.

What moves a rate is the profile of what you process, not just the total. Providers price on average transaction value, chargeback rate, refund rate, the mix of domestic and international cards, and how predictable your volume is. A store with a low dispute rate and a high average order value is a materially better risk than one with the same monthly total spread across small, frequently refunded orders.

That means the preparation matters more than the ask. Arriving with your own figures — processed volume, average transaction value, dispute rate, refund rate, domestic share — turns the conversation into a pricing discussion. Arriving without them produces a standard-rate answer.

Two structural changes are worth pricing alongside the rate itself: reducing the fixed per-transaction component if your average order value is low, and settling in fewer currencies to cut conversion spread. Either can be worth more than a few basis points off the headline percentage.