Accepting Payments Online: Gateways, Fees and Setup

Every dollar of online revenue passes through a payment stack most owners never inspect. The right setup feels invisible; the wrong one quietly taxes every order, every renewal and every abandoned cart. On our illustrative home-fragrance brand — $1.2 million of online revenue across 18,000 orders a year, an average basket of $67 — the gap between a careless stack and a tuned one is worth tens of thousands a year.

This guide walks the chain: how the published rate cards from Stripe, PayPal and platform processors compare, what fees do at your volume, why checkout design moves more money than fee negotiation, and where subscriptions, fraud and integration fit. Every rate quoted comes from a published rate card or public research; every model is illustrative with assumptions shown.

Here is the short version before we go deep.

Key Highlights

• Published US card rates: Stripe 2.9% + $0.30 and PayPal Checkout 3.49% + $0.49 — a $67 basket costs $2.24 via Stripe versus $2.83 via PayPal.
• Across the brand's 18,000 orders, that is about $40,400 a year via Stripe versus roughly $50,900 through PayPal Checkout — a $10,500 gap for identical sales.
• The fixed per-order fee dominates small baskets: effective cost runs 4.10% at a $25 basket but 3.02% at $250 (Stripe's published rates).
• 70% is the long-run average cart-abandonment rate (Baymard Institute); surprise shipping and fees are the top cited reason — 48% (rounded).
• Lifting checkout completion from 30% to 33% adds about $120,000 a year at this traffic (illustrative) — roughly ten times the fee gap.
• Recurly's published benchmarks put involuntary churn near 1% of subscribers monthly; dunning left 88 extra subscribers by month twelve in our model — about $3,300 of MRR.
• Card fraud cost merchants about $33 billion in 2022 (Nilson Report); each dispute carries a $15–$20 published fee plus goods and staff time.
• Time to first live payment runs from 2 days with payment links to about 70 days for a full custom API (illustrative agency ranges).

The Stack Behind Every Online Payment

When a customer clicks buy, the request makes four hops: the checkout captures details, the gateway encrypts and routes them, the processor talks to the card networks, and an acquiring bank settles money into your account. Most providers collapse these roles into one product with one rate. What you are choosing is a rate card, a checkout and an integration surface.

If you sell through Shopify, most of this is settled for you: Shopify Payments matches Stripe's published 2.9% + $0.30 online rate on the Basic US plan, and a third-party gateway triggers a 0.5–2.0% surcharge (both published). WooCommerce and custom stacks leave the choice open — that is where this guide applies. Either way, the rate card is only the headline.

Key takeaway: You are not picking a payment company; you are picking a rate card, a checkout and an integration surface at once. Judge all three together, because each taxes your revenue in a different currency — basis points, conversions or engineering time.

Factor 1: Gateway Comparison — What the Published Rate Cards Say

The major US providers publish their card rates, and the differences look tiny until you multiply them by an order count. The table puts the published cards side by side, with what each means for the illustrative $67 basket.

PayPal's advanced card processing looks cheapest per order, but it requires PayPal's full commerce-platform integration — and the pricier wallet option exists for a reason: many buyers complete inside PayPal who would abandon a card form. That is why most stores run cards and wallets side by side.

Two details matter more than the headline percentages. Stripe's US pricing adds 1.5% for international cards and 1% for currency conversion, so cross-border mix quietly raises your effective rate. Legacy merchant accounts quote "interchange-plus" deals that can beat flat rates at scale but add monthly fees, statement fees and a contract.

Key takeaway: Rate cards differ by fractions of a percent, and fractions never decide anything by themselves. Compare effective cost per order at your real average basket, then weigh wallet reach and integration effort on top of the arithmetic.

Factor 2: Fee Math at Volume — Why AOV and Mix Beat the Headline Rate

The fixed fee is why two stores on the same rate card pay very different percentages. On Stripe's published 2.9% + $0.30, a $25 basket costs 4.10% effective while a $250 basket costs 3.02% — the chart above shows the same bend for PayPal. At the brand's $67 average, effective cost is 3.35%, or $2.24 per order.

Multiplied out, the annual number gets executive attention: 18,000 orders at $2.24 is about $40,400 a year via Stripe, versus roughly $50,900 through PayPal Checkout — a $10,500 gap for the same sales. Mix moves it further: if 12% of orders run on international cards, the published 1.5% surcharge adds about $2,200 a year. The point is not dropping PayPal; it is steering mix.

Two levers change the math at volume. First, ACH bank debit for B2B or high-ticket invoices: on a $1,800 invoice, card fees run about $52.50 while ACH caps at $5 published — 200 such invoices a year saves roughly $9,500 (illustrative math). Second, custom pricing: Stripe publishes custom interchange-plus pricing for larger businesses — historically those above roughly $80,000 a month — so at about $1 million a year online, a negotiation is worth a conversation.

Key takeaway: Effective rate equals the headline rate plus the fixed fee spread over your basket size. Raise average order value, route B2B volume to ACH, and revisit custom pricing near $1 million a year — in that order.

Factor 3: Checkout UX — Where Conversion Is Won or Lost

Baymard Institute's long-running research puts the average cart-abandonment rate near 70%, and its 2024 survey of why shoppers walk away reads like a checkout to-do list (values rounded; respondents could give multiple reasons). Every bar below is a product decision, not a fee schedule.

The fixes are unglamorous and mostly free: show total cost up front, offer guest checkout, place express wallets above the form, autocomplete addresses, write payment errors in plain language, and support the methods your audience expects. None requires custom code on a hosted checkout — it requires someone owning the funnel and measuring completion weekly.

The stakes dwarf fee arithmetic. If the illustrative brand lifts checkout completion from 30% to 33% — three points, very achievable with express wallets and transparent shipping — that is about 1,800 extra orders, roughly $120,000 a year at the same traffic (illustrative). That is ten times the annual Stripe-versus-PayPal gap. When checkout speed itself is the bottleneck, going headless can pay for itself: the headless e-commerce storefront case study shows the before-and-after, and our conversion rate optimization guide covers the testing method.

Key takeaway: A basis point on the rate card is noise; a point of checkout completion is a business. Fix the abandonment reasons Baymard names before you spend a single meeting negotiating basis points.

Factor 4: Subscriptions — Dunning Pays for Itself

Recurring revenue changes the picture again: the same 2.9% + $0.30 applies to every renewal, but a new failure mode appears. Involuntary churn — cards expiring, being reissued, or declining — churns roughly 1% of subscribers per month, according to Recurly's published benchmarks.

The fix is dunning: automatic retries on a smart schedule, network account updaters that refresh changed card details, and pre-dunning emails that nudge customers to fix a card before cancellation. Take an illustrative meal-prep subscription running 2,000 subscribers at $38 a month: 1% involuntary churn is about 20 failed payments a month, roughly $760 at risk; recovering about half is $380 a month back immediately — and the compounding effect is larger, as the chart shows.

In the model, shaving monthly churn from 3.0% to 2.5% leaves about 88 more subscribers at month twelve — roughly $3,300 of monthly recurring revenue (illustrative). Dunning tooling on the major gateways costs a small fraction of that, which makes it one of the highest-ROI lines in the stack. Bookings businesses get a cousin of the same benefit from card-on-file deposits — our retreat booking engine case study shows the pattern.

Key takeaway: Failed renewals are a plumbing problem with compounding returns. Turn on retries, account updaters and pre-dunning emails before you spend another dollar on acquisition.

Factor 5: Fraud and Disputes — Cheap Rules Beat Cheap Rates

Card fraud cost businesses about $33 billion in 2022, according to the Nilson Report, and every online store pays a share. The direct costs are published and look small: $15 per dispute on Stripe and $20 on PayPal (US rate cards). The honest all-in cost per dispute is the fee plus lost goods plus 20–40 minutes of staff time — on the illustrative brand, about $107 each.

At a 0.3% dispute rate — 54 disputes across 18,000 orders — that is about $5,800 a year (illustrative). Basic controls cut it roughly in half for an hour of configuration: CVC and AVS matching, velocity rules that flag card-testing bursts, a blocklist, and tracking on every shipment. European traffic adds a gift: PSD2's Strong Customer Authentication shifts liability to the issuer when 3-D Secure runs, so enabling 3DS on EU orders is usually free protection. Watch your ratio against the networks' public thresholds — Visa's monitoring line sits at 0.9%.

Key takeaway: Fraud losses are a rules problem before they are a rates problem. Configure the free controls, scope 3DS to where regulation rewards it, and treat every dispute as free training data for your rules.

Factor 6: Integration Options — Hosted, Drop-in or Full API

The last decision is architectural, and it trades control against speed. Major gateways offer a ladder of integration routes, each with a different time-to-live, PCI compliance scope and degree of UX control.

The hosted end of the ladder keeps you in the lightest PCI self-assessment (SAQ A) because card data never touches your servers; the full-API end buys control at the price of a compliance program. Most businesses overestimate how much custom they need — hosted checkouts now handle wallets, saved cards and subscriptions out of the box. We build fully custom flows when the payment experience is the product, as in our fintech wallet case study, and hosted or drop-in for everything else, including the platform choices in our e-commerce platform guide.

Key takeaway: Start at the top of the integration ladder and step down only when a hard requirement forces you. Speed to first revenue beats pixel-level control almost every time.

What Your Business Should Do

Days 1–30: Audit what you have.

• Compute your effective rate from last quarter's processor statement: total fees divided by total volume (the illustrative brand's landed at 3.35%).
• Buy something on your own store from a phone; count taps, hunt for surprise costs and forced account creation.
• List your dispute rate, top decline reasons and — if you bill recurring — your involuntary churn for the last 90 days.

Days 31–60: Fix the funnel before the fees.

• Switch on express wallets and guest checkout, and show total cost before the final step.
• Turn on the free fraud controls — CVC and AVS matching, velocity rules — and name a dispute-response owner.
• If you run subscriptions, enable smart retries, account updaters and pre-dunning emails; measure recoveries weekly.

Days 61–90: Optimize the rails.

• Steer the mix: ACH for B2B and high-ticket invoices, cards on the cheaper rail, wallets where they convert.
• Model custom pricing against your effective rate if you process near $80,000 a month.
• Set a standing quarterly review — effective rate, checkout completion, dispute rate, involuntary churn — on one dashboard.

The Bottom Line

Payments is one of the few parts of an online business where small, boring changes compound into five-figure annual differences: the right rail per transaction, a checkout that respects the buyer, renewals that heal themselves. The math here is deliberately simple — rerun it with your own volume and basket size, and the priority list usually reorders within an hour.

If you want a second pair of eyes on your stack, Book a free strategy call and we will walk your numbers with you — or send us a message with your questions and we will answer within one business day.