Shopify Migration Launch QA Checklist: Go/No-Go Blockers
Shopify migration launch QA checklist covering release blockers, redirects, pricing, stock, analytics and sign-off for U.S. DTC and B2B merchants.

A store can accept an order and still be unlaunchable. A successful checkout doesn't prove that the customer received contractual pricing, the warehouse received the order, or marketing recorded the purchase correctly. Shugert Marketing's launch QA therefore covers customer events, ERP and 3PL integrations, checkout, and order validation, not just storefront appearance.
For established DTC and B2B merchants moving from WooCommerce, Magento, BigCommerce, or a headless build, a Shopify migration launch QA checklist should function as a release contract. Each business-critical behavior needs an owner, observable evidence, and a pass condition before cutover. For migrations with meaningful organic traffic, pricing complexity, or ERP and 3PL dependencies, see how Shugert scopes Shopify migration engineering and launch accountability.
Planning a replatform? Our Shopify migration service moves products, customers, orders and redirects as one SEO-safe scope.
Table of Contents
- Launch Blockers That Stop a Cutover
- Dependency Register Across Operations, Finance, Marketing and Support
- How to Test Redirects Before DNS Cutover
- Reconciling Records and Walking the Customer Journey
- Purchase Event Checks Before Launch
- Performance Gate Using Core Web Vitals
- Confirm Go or No-Go Before the Cutover Window
Launch Blockers That Stop a Cutover
Hidden pricing rules deserve an explicit launch gate. A product page can display the right price while quantity breaks, account-level discounts, or automatic promotions produce the wrong checkout total.
Launch blockers and pre-launch pass tests
| Blocker | Why it stops cutover | Pass test before launch |
|---|---|---|
| Hidden pricing rules | Tiered, quantity, or account rules are missing or applied twice | Each rule has a Shopify equivalent or documented replacement, tested at cart and checkout |
| Failed payments | Customers cannot complete orders, or payment capture isn't live | Test and authorized live payment paths complete and reconcile to an order |
| Wrong contractual pricing | Customer or B2B price lists differ from the signed agreement | Sampled SKUs match the approved price file in every applicable market and customer group |
| Material stock differences | Storefront availability differs from warehouse or ERP records | Variance stays within the tolerance approved by the operations owner |
| Stale product feeds | Ads, marketplace, or ERP feeds reference legacy URLs or old prices | Feeds are regenerated, validated, and pointing to new Shopify URLs |
| Missing order-routing dependencies | Orders don't reach the 3PL, ERP, or tax service | Test orders route to the correct fulfillment location and return status |
| Lost or duplicated orders | Orders drop or import twice during cutover | Order counts and totals reconcile against the source system for the agreed cutover scope |
| Critical redirect or indexing defects | Priority legacy URLs return errors, chains, or targets marked noindex | Priority URLs resolve to relevant product, collection, or editorial destinations, with a successful final response and correct indexability |
A blocker stops launch until it passes a recorded retest. Any exception requires written risk acceptance by the accountable merchant owner and engineering lead; it must not waive an unworkable checkout, material order loss, or unreconciled payment and inventory failure. Noncritical defects can enter the post-launch backlog when they do not threaten contractual obligations, fulfillment, payment, measurement, or critical search access.
Cosmetic spacing, minor copy errors, and scripts that are slower than preferred but still pass the agreed performance standards generally don't justify postponement. The distinction is commercial impact, not how conspicuous the issue looks on a screen.
Release rule: A polished homepage isn't evidence that the business can trade correctly.
The go/no-go register is the release record containing each issue's evidence, severity, owner, and retest status. Evidence should identify the environment, customer context, expected result, and actual result, with an order reference, crawl output, or screenshot where appropriate.
Every blocker belongs in that register before the cutover meeting. A verbal assurance that engineering is handling it isn't a passing test, and a fixed ticket isn't closed until its owner records the retest.
Dependency Register Across Operations, Finance, Marketing and Support
The dependency register prevents teams from assuming someone else owns a connection. Engineering may configure an integration, but finance must approve its pricing behavior, operations must approve its stock and fulfillment behavior, and support must confirm that agents can serve customers.
Each entry needs a named person, not only a department. The template below uses departments as starting points; the working version should identify the individual authorized to accept the result.
Dependency register template with example entries
| Dependency | Owner | Target replacement | Acceptance test |
|---|---|---|---|
| Legacy pricing rules | Finance | Supported Shopify B2B or catalog pricing configuration, a vetted pricing app, or an approved custom integration | Sampled prices match the approved price file |
| ERP stock feed | Operations | Native sync or middleware | Storefront stock matches the ERP count within the approved tolerance |
| Shopping and ads product feed | Marketing | Feed generated from Shopify | Feed validates and contains no legacy URLs |
| 3PL order routing | Operations | Fulfillment connector or app | Test order routes to the correct location with status returned |
| Tax and shipping rules | Finance | Shopify tax and shipping settings rebuilt | Test carts match expected tax and rates by zone |
| Helpdesk access to orders | Support | Helpdesk connected to Shopify order data | An agent finds a test order and its fulfillment status |
| Consent banner and tag manager | Marketing | Consent tool configured for the new storefront | Tags respect each consent state during testing |
Target replacement should name the actual configuration, app, middleware service, or custom component. Writing Shopify beside every dependency conceals the decisions that still need engineering work.
Native settings reduce moving parts when they reproduce the required behavior. An app can shorten implementation, but it adds permissions, configuration, and vendor dependencies. Middleware or custom code may be necessary when the ERP owns inventory or contractual logic, but someone must own retries, alerts, and maintenance.
Make acceptance understandable outside engineering
Finance shouldn't need to interpret an API response to approve contractual pricing. Its evidence can be an approved price file beside the resulting customer-specific cart. Support can demonstrate order lookup through the interface agents will use.
For asynchronous connections, acceptance must include the return path. A connector accepting an order doesn't prove that fulfillment status returns to Shopify or becomes visible to support.
- Pricing ownership: Finance approves the hidden-pricing-rules and contractual-pricing pass tests.
- Operational ownership: Operations accepts stock tolerances and order-routing outcomes.
- Marketing ownership: Marketing approves feed destinations and consent behavior.
- Support ownership: Support confirms access using an agent account, not an administrator session.
Any required dependency without an owner or passing acceptance test remains a blocker. Shugert Marketing assigns one senior team responsibility for storefront implementation, redirects, analytics validation, and launch QA, while merchant teams retain business acceptance. The broader Shopify migration planning guide covers scope decisions beyond release readiness.
How to Test Redirects Before DNS Cutover
Redirect testing belongs before the domain switch, not after marketing notices broken landing pages. The release evidence must come from the redirect configuration that will serve production traffic. For stores with established rankings or many legacy URLs, Shugert's technical SEO migration protection covers URL mapping, indexation checks, and monitored releases.
Run the mapped URL inventory
For native store redirects, verify Shopify's URL redirect rules and path limitations before approving the import. The exact test method depends on whether redirects live in Shopify, a reverse proxy, or another routing layer.
Build the legacy inventory. Combine a full crawl with analytics landing pages, Search Console exports, backlink destinations, and campaign URLs. Group product, collection or category, editorial, account, and campaign paths so a missing page family doesn't disappear inside a large spreadsheet.
Map each legacy URL to its intended Shopify destination. Record deliberate retirements with a reason instead of sending unrelated pages to the homepage. The destination should preserve the visitor's intent, not merely avoid an error.
Exercise the production redirect layer before public cutover. Google's site-move guidance recommends server-side permanent redirects such as 301 or 308 and retaining migration redirects as long as possible, generally at least one year. Use a controlled routing arrangement appropriate to the architecture, preserving the production hostname and TLS behavior; check direct resolution, chains, loops, accidental errors, and protocol or hostname variants.
Inspect the destination references. Canonicals, internal links, XML sitemaps, hreflang references, and structured data should use the intended Shopify URLs. A working redirect doesn't compensate for a target page that tells search engines another page is canonical.
Hypothetical tire-size QA example: A legacy URL for 205/55R16 should reach its matching size collection, not a general tire category. A neighboring size must not reach that same collection through an overly broad rule. Test the boundary because a plausible-looking collection can conceal an incorrect mapping.
Record the result against each URL. Preserve the requested path, response status, destination, final response, indexability, and tester. Attach critical failures to the go/no-go register rather than keeping them in an unreviewed crawl file.
Separate theme preview from routing proof
A staging-theme preview can validate templates and content. It can't prove production hostname routing, all store-level settings, or the redirect layer customers will encounter.
If production-equivalent routing cannot be exercised safely beforehand, record that limitation explicitly. Require configuration inspection before release and immediate live validation during the controlled cutover window.
For rule creation and import mechanics, use the 301 redirect implementation guide. Teams evaluating SEO tooling can consider The SEO Agent as an optional third-party example, but no content tool replaces per-URL migration evidence.
A redirect passes when the intended visitor reaches the intended, indexable destination, not merely when the browser stops reporting an error.
Reconciling Records and Walking the Customer Journey
Begin with the approved migration scope, not an expectation that every legacy count must remain identical. Reconcile products, variants, customers, collections, and other approved entities, including exclusions for discontinued SKUs, duplicate records, and deliberately retired pages.
Counts establish completeness within that scope. Field-level checks establish whether the migrated records can support sales.
Follow a representative order to fulfillment
Hypothetical journey: A returning wholesale customer follows a legacy product link, reaches the intended indexable Shopify product or collection, signs in, and selects a variant. The tester validates SKU, dimensions, customer price, available stock location, and collection membership before adding anything to the cart.
The journey then continues through quantity changes, account permissions, shipping rates, tax, payment, and fulfillment routing. Operations verifies the destination warehouse, while support confirms that the order and subsequent status are accessible through its normal tools.
This test should favor commercially important and structurally difficult products:
- Revenue exposure: Products and landing pages that contribute materially to sales.
- Catalog complexity: Products with unusual variants, dimensions, bundles, or location rules.
- Customer differences: Retail, wholesale, contract-priced, and localized journeys where applicable.
- Operational exceptions: Unavailable stock, restricted shipping destinations, and split-location fulfillment.
The cutover delta import needs its own reconciliation. New orders, changed inventory, edited prices, and newly created customers can invalidate an otherwise successful rehearsal.
Use catalog evidence without extending the claims
The published Llantas Cavazos migration case study documents 8,000+ active SKUs, 84 size collections, tire-dimension organization, URL mapping, and legacy Merchant Center cleanup. That supports a launch test centered on dimensional catalog accuracy and destination mapping, not an assumption about vehicle-fitment functionality.
It also illustrates why feed cleanup and storefront checks belong in the same acceptance process. A correct collection is commercially insufficient if the shopping feed still advertises an obsolete destination or price.
Shopify announced support for 2,048 variants per product in its product-variant update. This platform limit does not itself guarantee support in every theme, app, API client, or downstream system. Validate the selected integrations against the intended product structure.
The announcement alone isn't proof that an ERP connector can process the most complex product. Test creation, pricing, images, inventory synchronization, and exported payloads before accepting the architecture. Record the actual source-to-target field mapping and reconcile imported values before signing off.
Purchase Event Checks Before Launch
Purchase event test paths
| Test path | What to inspect | Pass condition |
|---|---|---|
| Standard purchase | Order ID, currency, value, and purchase event in each destination | A single logical purchase per destination, with value matching the Shopify order under the approved revenue definition |
| Thank-you page revisit | Purchase behavior on reload or return | No additional purchase is recorded |
| Payment failure | Events associated with a declined payment | No purchase is recorded for the failed attempt |
| Consent denied | Tag behavior under each consent choice | Only destinations permitted by the approved consent configuration receive events |
| Discount code order | Revenue value after the discount | Value matches the net Shopify order total under the agreed tax and shipping treatment |
| Refund or cancellation | Downstream adjustment behavior | Refund handling matches the approved reporting rule |
A pixel firing isn't sufficient evidence. Shopify's pixel migration guidance calls for expected events, equivalent implementation logic, events firing only once, comparisons within the same analytics service, and matching privacy and consent settings.
The acceptance test must therefore preserve event meaning, deduplication, attribution, consent behavior, and revenue accuracy. A dashboard can show purchases while recording the wrong value or crediting the wrong channel.
Make every test order traceable
For each controlled order, log its Shopify order ID, transaction ID, currency, value, consent state, customer journey, and expected destinations. Inspect browser network requests and the applicable platform debug tools, then reconcile the resulting transaction IDs against Shopify orders.
Where browser and server tracking both operate, verify the destination's supported deduplication mechanism. Matching amounts aren't enough; the implementation must identify both deliveries as the same logical transaction.
Maintain a pre-launch and post-launch event dictionary with:
- Event meaning: The customer action that qualifies, including variant selection and checkout initiation.
- Payload rules: Product identifiers, currency, revenue treatment, and transaction identifiers.
- Delivery ownership: Native integration, app pixel, custom pixel, or server-side integration.
- Consent and attribution: Permitted destinations and expected treatment of paid, organic, direct, and email journeys.
Failure testing should distinguish a genuine decline from a broken integration. Spell's card decline explainer provides context for selecting hard- and soft-decline scenarios; the QA record still needs the actual gateway result and observed analytics behavior.
Inventory legacy theme scripts, customer-event integrations, and checkout-related tracking separately. Copying old code into the new implementation without identifying its owner can create competing purchase emitters.
Measurement gate: Marketing approves the recorded transaction, not just the presence of a tag.
Test-mode orders aren't guaranteed to reach every marketing integration. Where required, schedule authorized production validation with a real low-value order, approved payment handling, and an agreed refund or cancellation procedure.
Performance Gate Using Core Web Vitals
Core Web Vitals launch thresholds
The following bands follow Google's Core Web Vitals guidance.
| Metric | Good | Needs improvement | Poor |
|---|---|---|---|
| Largest Contentful Paint (LCP) | 2.5 seconds or less | Above 2.5 through 4 seconds | Over 4 seconds |
| Interaction to Next Paint (INP) | 200 milliseconds or less | Above 200 through 500 milliseconds | Over 500 milliseconds |
| Cumulative Layout Shift (CLS) | 0.1 or less | Above 0.1 through 0.25 | Over 0.25 |
Google assesses Core Web Vitals using the 75th percentile of real-user page visits, with mobile and desktop evaluated separately where data is available. These thresholds are useful launch criteria, but a prelaunch lab run does not establish production field performance. The release record should distinguish historical field measurements from controlled lab diagnostics and identify where new production evidence is still unavailable.
Capture the pre-migration baseline for the homepage, product detail pages, collection pages, cart, search, and highest-revenue landing pages. Compare equivalent templates on mobile and desktop under representative device and network conditions, preserving URLs, test conditions, and results.
Test the loaded storefront, not the unfinished theme
Consent banners, third-party apps, analytics tags, reviews, personalization, chat, and payment elements must be present during acceptance testing. Testing a stripped-down theme and enabling these components afterward invalidates the comparison.
Include consent-accepted and consent-denied states. Exercise product selection, filters, cart updates, and other meaningful interactions rather than measuring only the initial load.
Assign fixes according to the cause:
- Slow LCP: Inspect hero-image dimensions and priority, Liquid rendering, font loading, and blocking scripts. The theme or media owner should receive the relevant trace.
- Poor INP: Profile interaction handlers and app or pixel execution. Identify the component responsible before removing functionality.
- High CLS: Reserve space for images, recommendations, reviews, and promotional content. The layout must remain stable when delayed components arrive.
Removing an app can improve performance but break merchandising, consent, or support workflows. The dependency owner must approve that trade-off and rerun the affected acceptance test.
Preserve page-level measurements rather than reporting only a blended score. The Core Web Vitals benchmark resource supports interpretation, while the release decision should use the merchant's agreed thresholds and documented regressions. Identical scores after launch aren't a credible promise.
Confirm Go or No-Go Before the Cutover Window
Release only when non-waivable blockers pass retest and any remaining exceptions are explicitly accepted in writing. Merchant operations approves business workflows. Engineering approves data, integrations, and release recovery; SEO and analytics owners approve their respective checks.
Execute a controlled release
Shopify's Plus launch checklist recommends lowering DNS TTL to 300 seconds, or five minutes, approximately one week before launch. This reduces cache duration; it doesn't guarantee simultaneous propagation across every resolver.
The release owner coordinates the operational sequence:
- Freeze and reconcile: Control source-platform changes, complete delta imports, and verify the cutover order ledger.
- Switch deliberately: Update DNS and the primary domain during the approved window, then disable Shopify Payments test mode before authorized live payment validation.
- Validate immediately: Crawl critical legacy URLs and confirm checkout, payment capture, order export, inventory, tax, and shipping behavior across representative networks and devices.
Define recovery before changing traffic
Failed payments, lost or duplicated orders, and critical redirect or indexing defects trigger an immediate recovery assessment. The named engineering release owner calls rollback in coordination with merchant operations.
The runbook must define the rollback window by operational conditions: the source remains usable, both destinations' orders can be reconciled, and integrations can be restored safely. DNS reversal isn't data reversal. Reduced TTL can shorten cache-related recovery delay, but it doesn't establish a guaranteed rollback deadline.
Define a post-launch monitoring period, often including checkpoints across the first 30 days, with named owners and escalation thresholds. Actual duration, coverage, and cadence must be included in the agreed scope rather than implied as a universal service guarantee. Launch-critical checks cover checkout failures, order-export exceptions, stock discrepancies, and redirect errors, with before/after evidence and handoff. Identical legacy counts and zero SEO fluctuation are not realistic promises.
For Shopify Plus, B2B, ERP or 3PL integrations, subscriptions, or a large legacy URL footprint, agency involvement may be justified when cutover failure would be materially costly. Shugert's custom Shopify development addresses the integration and storefront engineering that must pass these acceptance tests.
Request a migration scope with the source platform, critical integrations, and target launch window. Full-service Shopify migrations typically start at $5,000, with timelines of 4 to 24 weeks depending on scope. Smaller fixed-scope engineering projects are quoted separately; the full Shopify migration service scope explains the work covered by an approved launch plan.
Migrate without losing rankings
Llantas Cavazos moved 8,000+ SKUs in under 7.5 weeks and kept 90%+ of organic traffic in week one. Read the case study
Most quotes returned within 1 business day