Legacy checkout scripts carried years of quiet customizations: gift messages, delivery date pickers, warning banners. Moving to checkout functions and blocks surfaces the ones nobody documented. Inventory first, migrate second.
Every long-lived Plus store has them: scripts installed for a promotion in 2021 that quietly became load-bearing. A line that adds a gift message field. A rule that hides a shipping method for PO Boxes. Nobody remembers writing them, and the migration discovers them one customer complaint at a time.
Inventory behavior, not files
Read each script and translate it into a behavior statement: what the customer or the order experiences because this exists. 'Adds delivery date picker for local delivery orders' is a behavior. The raw script is not. Most stores find behaviors nobody can explain, which is the point of the exercise: unexplained behavior that customers rely on is still behavior you must replace or consciously retire.
Test the live checkout as each persona the scripts serve: gift buyer, PO Box customer, wholesale account, international address. Scripts usually condition on exactly those traits, and a plain vanilla test order sails through a migration that breaks every edge case.
The usual casualties
Gift message and gift wrap options lead the list, because they lived in scripts on many stores and their absence is only noticed by the recipient. Field validations come next, especially the kind that warn rather than block, which functions-based replacements often implement as hard errors or not at all.
Order tagging and post-purchase routing rules break silently in the other direction: nothing looks wrong at checkout, but the warehouse loses the tag that routed gift orders to special packing. Check the order in admin after each test, not just the checkout screens. The script's real job is often downstream.
Map each behavior to its new home
Checkout blocks cover display customizations. Checkout functions cover discounts, shipping and payment rules, and validation. Some script behaviors split across both, and a few, mostly anything that reached outside checkout's data, have no equivalent and need an app or a process change.
Write the mapping down as a table with three columns: behavior, replacement, verified. Verified means a test order proved it, not that the replacement exists. Migrations fail on the gap between installed and verified more than on any missing feature.
Cut over with a watch window
Migrate in a low-traffic window and watch three things for the first days: checkout completion by segment, support contacts mentioning checkout features, and order-note or tag patterns in admin. The segments matter because script breakage is segment-shaped; overall completion can look fine while gift orders quietly fall apart.
Keep the old scripts documented and the rollback decision made in advance: which signal, at what threshold, sends you back. A migration with a decided rollback rule is a test. A migration without one is a hope.
Related reading
These earlier guides go deeper on connected questions: One Page vs Three Page Checkout: How to Test the Layout Honestly, Checkout Extensibility Upgrade Testing: Moving Apps Without Breaking Orders, and Tax Display at Checkout: When Shoppers See the Full Price Matters.
Frequently asked questions
How do I find out what my checkout scripts actually do?
Read each one and write a plain-language behavior statement, then verify it in the live checkout with test orders for each customer type the script might condition on. Behaviors nobody can explain get flagged for replacement or conscious retirement.
What most commonly breaks in migration?
Gift options, soft validation warnings, and downstream order tagging. The first two are visible at checkout; tagging breakage is silent and surfaces in fulfillment, so check test orders in admin, not just the checkout flow.
Can every script behavior be replaced with blocks or functions?
Most can. Display changes map to blocks; discount, delivery, payment, and validation logic maps to functions. Behaviors that depended on data outside checkout may need an app or a workflow change instead.
When should I do the migration?
In your lowest-traffic window, after the behavior inventory and mapping are complete, with a pre-agreed rollback threshold. Avoid peak promotions; script breakage hits hardest exactly when edge-case orders spike.