There is a moment in every Shopify Plus brand's life where someone in a finance meeting works out what the brand is paying in processing fees and surcharges, and the conversation pivots from product to payments. The next question is always the same: what would it actually take to replace Shopify's checkout with our own?

The short answer: it is faster than people think, cheaper than people expect, and harder than the demos make it look.

This post is the honest version. I will tell you upfront — for most brands, the answer is "do not". The brands where it makes sense know exactly why they are doing it and have the engineering culture to back it up. If you are not in that group, the savings will not survive contact with the conversion hit you take from doing it badly.

What "Stripe-native" means

A Stripe-native checkout is a checkout flow built directly on Stripe's APIs — usually the Payment Element or Checkout Sessions — running on your own infrastructure rather than Shopify's checkout pages.

It still uses Stripe's PCI-compliant card capture. It still uses Stripe Radar for fraud. It still uses Stripe Tax for GST. The card details never touch your servers. The difference is that the page wrapping all of that is yours — your design, your code, your domain, your conversion data, your A/B testing infrastructure.

For brands on Shopify Plus, this means either:

Replacing Shopify entirely with a custom Saleor or Medusa backend
Keeping Shopify products and admin, but routing the actual payment through Stripe directly via a custom checkout layer

Option 2 is technically tricky on Plus because Shopify wants checkout to live on its infrastructure. Checkout Extensibility gives you some surface for customisation, but the checkout itself is still Shopify's. Going fully Stripe-native usually means moving the commerce backend off Shopify too.

The case for staying on Shopify Checkout

Before I make the case against, the case for.

Shopify checkout is extremely well-tuned

Shopify has spent the better part of a decade optimising their checkout flow at scale. Shop Pay, Apple Pay, Google Pay, address autofill, returning-customer recognition — all defaults. Their published numbers suggest Shop Pay alone lifts conversion by up to 15% for returning customers. Independent attribution suggests the real number is 5% to 10%, but it is a real lift.

If you have not built the resources to ruthlessly optimise your own checkout, Shop Pay is genuinely hard to beat. Most custom checkouts I have seen built by small teams underperform the platform default for the first 18 months.

One PCI compliance scope

Shopify takes on all the PCI obligations. You take on essentially none. Going custom means your team has to think about PCI SAQ-A scope (the easiest tier, possible if you use Stripe Elements correctly) and document it. Not difficult, but not free.

One bug-fix surface

If Shop Pay breaks, Shopify fixes it. If your custom Stripe checkout breaks at 11pm on a Thursday because Stripe pushed a deprecation, that is your team's problem. Most teams do not appreciate how much of their checkout's reliability is currently outsourced until they own the surface.

What the savings actually look like

Now the math. This is where the decision gets interesting.

Processing savings

Shopify Payments on Plus is roughly:

Domestic Australia: ~1.6% to 1.75% + fixed fee
International: ~3.1% + fixed fee

Stripe Australia direct is:

Domestic Australia: 1.75% + A$0.30 (1.925% + A$0.33 after GST)
International: 3.25% + A$0.30 (before GST)

At the headline rates, Shopify Payments looks similar or slightly cheaper. The catch is the Plus surcharge on third-party gateways — that 0.20% extra that some Plus brands pay because they have a multi-gateway setup — disappears when you leave Plus entirely.

The bigger lever is volume-based negotiation. Stripe's published rates are the starting point. At volume above $2M AUD/year processed, you can negotiate. At $10M+, brands routinely land interchange-plus pricing that is meaningfully below 1.6% blended for domestic Australian cards. The processing margin recovery for a brand doing $10M is in the $40,000 to $80,000 AUD/year range.

That is real money. It is also the smallest part of the total saving.

Platform fee savings

Shopify Plus base is $2,300 to $2,500 USD/month. Above $800k USD monthly sales, it becomes 0.35% to 0.40% of revenue. For a $10M AUD brand: roughly $40,000 to $50,000 AUD/year in platform fees.

A custom Saleor or Medusa stack hosted on AWS Sydney runs $2,000 to $5,000 AUD/month all-in for infrastructure. Annual: $24,000 to $60,000.

Net saving: roughly nothing on infrastructure, but you remove the per-revenue scaling. At $20M revenue the Plus fee is $80,000+; the custom infra is still $30,000 to $60,000. The bigger the brand gets, the bigger the gap.

App stack savings

This is the largest line item that nobody costs properly upfront.

A Plus brand running 20 paid apps at an average $58 to $102/month is paying $14,000 to $24,000 AUD/year in app subscriptions. A custom build rolls a meaningful number of those apps into the codebase — upsell logic, bundle pricing, custom shipping rules, subscription dunning, gift cards, store credit, B2B portal features.

Realistic app subscription saving on a serious custom build: $8,000 to $20,000 AUD/year. Not huge. But it compounds.

The total picture for a $10M brand

Processing recovery: $40,000 to $80,000
Platform fee: $50,000
App stack: $15,000
Total saving: $105,000 to $145,000 AUD/year

Against a build cost of $150,000 to $250,000 AUD and an engineering retainer of $120,000 to $200,000/year, the payback is real but not fast. Two-year payback is the realistic case for a brand at this size.

The payback only works if conversion does not drop.

The hard part: not breaking conversion

This is what the demos do not show. Conversion-rate impact is the variable that turns this decision from a numbers game into an engineering question.

What a bad custom checkout does

A poorly-built custom checkout loses 2% to 5% of conversion versus Shopify's optimised checkout. On a $10M brand, a 3% conversion drop is $300,000 AUD/year in lost revenue. That eats the entire saving.

The places a bad custom checkout loses:

No Apple Pay or Google Pay on the first step
No address autofill (or autofill that does not work on AusPost addresses with unit numbers)
No saved-payment recognition for returning customers
A two-page flow instead of a single page
Validation errors that show up on submit instead of inline
Card declines that do not surface a useful error message
A 2-second checkout load time instead of a 200ms one
No discount-code field, or a discount-code field that takes 3 seconds to validate

Every one of these is a 0.3% to 0.8% conversion hit. Stacked up, you have given back the whole saving.

What a good custom checkout does

The brands who do this well start with the same conversion benchmark as Shopify. Sometimes they beat it. The pattern that works:

Single-page checkout. No "next" buttons.
Apple Pay, Google Pay, Link by Stripe all visible above the fold
Australia Post address autocomplete via AusPost AddressChecker or Loqate
Stripe Link for returning customers (their version of Shop Pay)
Inline validation, not submit-validation
Edge-rendered checkout (server-side rendered on Vercel, hydrated client-side) so the page is interactive in under 400ms
Apple/Google Pay express buttons on the product page and cart, not just checkout
A discount code field that validates in under 200ms via your own API
Real-time shipping rate calculation against AusPost / Aramex / Sendle APIs

This is doable. It is also a real engineering project. Six to ten weeks of focused work for a competent team. Not a checkbox.

Where Stripe Australia helps and where it does not

A few specifics worth knowing if you are scoping a build in 2026.

The Stripe Australian domestic rate of 1.75% + A$0.30 is the second-lowest Stripe rate in the world after the UK and EU, thanks to Reserve Bank of Australia interchange regulation that caps debit interchange at 0.20% and credit at 0.50%. Australia is a friendly market for custom Stripe checkouts.

Stripe Tax handles GST registration thresholds, location-based rates, and the ATO reporting requirement. You do not need a separate tax engine for an Australian-only brand.

Stripe Connect is not what you want unless you are a marketplace. For a normal DTC brand, you use standard Stripe accounts.

Where Stripe is weaker: no native subscription dunning that is as good as Recharge's. If you need pause/skip/swap flows for a subscription product, you either use Stripe Billing and build the dunning logic yourself, or you keep Recharge as a separate service alongside Stripe.

When to seriously consider it

I will give you the rule I use with brands who ask.

You should be building a custom Stripe checkout if all of these are true:

You are doing $8M+ AUD revenue
You have or are hiring at least one mid-level engineer for ongoing maintenance
Your conversion rate optimisation roadmap is currently blocked by Shopify's constraints
You have a 24-month time horizon and the cash flow to ride a one-quarter conversion learning curve
The processing and platform saving is at least $80,000 AUD/year by your own math

If two or fewer are true, stay on Shopify checkout. Use the time and money for inventory, ads, or product.

The honest bottom line

A Stripe-native custom checkout is technically straightforward and operationally significant. The hard part is not the build — it is matching the conversion performance of a platform that has spent ten years optimising the same flow.

The brands who succeed are the ones who treat the checkout as a permanent engineering investment. The brands who fail are the ones who treat it as a one-time project. There is no "ship it and move on" version of this.

A practical first step: run our free audit on your current product and cart pages. The report gives you the mobile load times, the SEO surface, and the specific friction points hurting conversion right now on Shopify. If the audit says your product pages are slow and your cart drawer is broken, those are cheaper wins than a Stripe rebuild — and you should fix them before deciding whether to leave the platform at all.

Stripe-Native Checkout: How Much Faster, Cheaper, Harder (2026)