Shopify checkout optimization guide (2026)

Shopify checkout optimization in 2026 means three jobs. Find the exact step where buyers drop, using Shopify's Conversion rate breakdown report and a GA4 checkout funnel. Fix settings that are plainly broken, like a missing express wallet or a required field nobody needs. Then A/B test the decisions that shape checkout before the buyer reaches it: shipping cost, delivery dates, the cart and payment messaging. Shopify locked its checkout code in August 2024, so most checkout wins now come from what you change upstream and measure on completed orders.

What you can change in Shopify checkout on each plan

Shopify switched off checkout.liquid for the information, shipping and payment steps on August 13, 2024. No store can inject scripts into checkout anymore. Every change goes through Checkout Extensibility, Shopify's app-and-settings framework, and your plan decides how much of it you get.

If you are on Advanced, the left column is your whole checkout toolkit. That sounds limiting until you look at where checkout abandonment starts. A buyer who sees a $12 shipping charge for the first time on the payment step decided to leave on the product page, when nothing told them it was coming.

Find the step where buyers drop before you change anything

Open the Conversion rate breakdown report in Shopify Analytics. It shows sessions, sessions with an add to cart, sessions that reached checkout, and sessions that converted, as one funnel. Pull the last 90 days and split it by device.

Then build a funnel exploration in GA4 with these events in order:

1. begin_checkout
2. add_shipping_info
3. add_payment_info
4. purchase

Shopify's Google & YouTube app sends begin_checkout, add_payment_info and purchase by default. Check that add_shipping_info shows up in your GA4 events before you lean on that step. Read the drop between each pair and match it to a likely cause.

Compare mobile and desktop before you trust any row. A store where 75% of sessions are on mobile can show a healthy desktop funnel and a broken mobile one, and the blended number hides both.

Your support inbox fills in the why. Search it and your post-purchase survey for "shipping", "delivery", "afterpay", "discount code" and "account". Twenty tickets asking when an order will arrive is a better hypothesis than any benchmark.

Fix the settings that are broken, not the ones you have opinions about

Some checkout problems are bugs. A wallet that is switched off, or a field that blocks orders, does not need a two-week test. Go through this list in Settings > Checkout and Settings > Payments:

• Shop Pay, Apple Pay and Google Pay are on, and the express buttons render on mobile Safari and Chrome
• Customer accounts are optional, or use sign-in links, so nobody has to create a password to pay
• Company name and address line 2 are hidden or optional
• Shipping address phone number is optional unless your carrier needs it
• Address autocomplete is on
• Every market you sell into shows local currency and a payment method buyers there use, such as iDEAL in the Netherlands or Klarna in Sweden
• Checkout loads in under three seconds on a mid-range Android phone over 4G
• Discount code field works with your top three active codes
• Abandoned checkout emails are switched on in Marketing > Automations

Anything past that list is a preference, and preferences are hypotheses. "Our buyers want a trust badge row above the pay button" might be true for a $300 skincare store with no reviews and false for a store with 4,000 reviews. The only way to know for your visitors is to test it.

The checkout layout is a good example. Shopify defaults to one-page checkout and lets every plan switch to three-page. It is a store-wide setting, so you cannot split traffic between the two. Leave the default unless your GA4 funnel shows a specific problem the other layout would solve.

Checkout test ideas that run before the checkout page

Because checkout code is locked, most checkout tests change something the buyer sees before checkout and judge it on what happens inside it. Each row below is a single hypothesis to test as one isolated change. None of them is a fix to apply to every store.

Pick the row that matches the biggest drop in your funnel and test that one first. If the shipping step leaks, start with shipping cost visibility. If carts never reach checkout, start with the cart drawer.

Plus stores can also test changes inside checkout. The pattern Shopify supports is to assign each visitor a variant before checkout, store it as a cart attribute, and have a checkout UI extension read that attribute to decide what to render. That needs a developer, and it still only measures well if the assignment happened upstream.

For more test ideas sorted by page type, see 30 ecommerce A/B testing ideas for Shopify.

How to A/B test checkout changes when checkout code is locked

The rule is simple. Randomise before checkout and measure on orders.

Split visitors on the product page or cart, where a testing script can still run. Then match each Shopify order back to the variant that visitor saw. An app with read access to orders can do that matching, so the result reflects real paid orders, refunds included, instead of a pixel firing on a thank you page.

Judge most checkout tests on revenue per visitor. A free shipping threshold test can lift conversion rate while cutting order value, and only revenue per visitor shows whether you came out ahead.

Traffic decides which tests are realistic. Take an example store with 40,000 monthly product page visitors, an 8% add to cart rate (3,200 cart sessions a month) and a 30% cart-to-order rate. At 95% confidence and 80% power:

That table changes the plan. At this traffic, bold changes like shipping cost visibility are worth testing, and a button colour in the cart drawer is not. Run your own numbers in the sample size calculator before launch.

Then run the test the way any A/B test should run:

1. Set the sample size before launch and do not check the result until you reach it.
2. Run for at least two full weeks, so weekday buyers, weekend buyers and payday buyers all land in both variants.
3. Call a winner at 95% significance and no lower.
4. Remove a losing or flat variant from the site and write down what it taught you.
5. Use that learning to write the next hypothesis.

Step 5 is where checkout work compounds. If showing shipping cost on the product page wins, the next hypothesis might test the threshold itself. If it loses, shipping cost probably was not the leak, and the next test moves to payment methods. A fixed list of 20 checkout fixes written in week one cannot react to either result.

For the full mechanics of splitting traffic on Shopify, see how to run Shopify A/B tests that prove revenue.

Who should run your checkout tests

Intelligems, Shoplift and Visually have done a lot to get Shopify stores testing, and Intelligems handles shipping and price tests well. The catch is the hours. Someone on your team still has to write the hypothesis, build the variant, QA it on every device, wait out the sample size and read the result. Our Shopify A/B testing tools guide compares what each app can test.

Retainer agencies take the work off your plate, and they get paid the same whether checkout completion moves or not. Most start from an audit and work through a fixed list of ideas for the quarter. We compared both routes in DIY vs agency A/B testing.

Cascayd is outcome-based. You add a pixel, give GA4 access and install a Shopify app with read access to orders. We research your funnel, design and build each test with AI assistance and a human checking every step, and send it to you in Slack with the reasoning. You approve it or ask for changes. One test runs at a time, and the next hypothesis comes from what the last one taught. Cascayd is paid only from a share of the measured uplift, so a flat checkout test costs you nothing. See how the Shopify CRO service works.

Your next step

Open the Conversion rate breakdown report in Shopify Analytics today and write down the single biggest drop on mobile. That number tells you which row of the test table to start with.

If you want that test built, approved in Slack and measured on real orders without adding hours to your week, book a call and tell us about your store.