A shipping-module change, cart fix or PrestaShop update may look correct while preventing orders in a less obvious scenario. Useful acceptance testing covers the purchase workflow, states the limits of the checks performed and leaves a recovery plan for problems after release. Here is a practical way to organise it without accidental production operations.
1. Define the change and the expected outcome
Write down what should behave differently and what must remain unchanged. Include the page, store and module versions, reproduction conditions and expected result. If the problem affects a wholesale customer, a particular currency or one multistore shop, the test needs those conditions too.
Identify neighbouring processes separately. A payment-module change may require cart, email and order-status checks. A product importer change may affect prices, images and variants. Regression coverage should reflect those dependencies rather than the number of modified files.
2. Prepare a current, isolated test environment
The test environment should reflect the store, theme, modules and settings relevant to the change. A copy several months out of date may hide a conflict present in production. Preserve any test-environment data you still need before refreshing it.
Separate the database, addresses and service connections. Check outgoing email, payments, shipping, scheduled tasks, webhooks and ERP integrations. A domain containing the word ‘test’ does not stop an application sending real messages or orders. Use demonstration data and limit access to the people conducting acceptance checks. Block search indexing as well.
Agree a safe way to test responses from external integrations. Do not connect a copy to the production ERP or shipping account simply to see whether it works. Sandbox settings and test credentials must apply to the connection actually being exercised.
3. Back up the store and define recovery
Before deployment, prepare consistent copies of files, the database and required configuration. Establish where they are stored, who can restore them and which point in time they represent. A successful backup notification does not replace checking the scope and ability to restore it.
Reverting files and restoring a database are different operations. After a schema change, older code may no longer work correctly. Restoring an older database may discard new orders. The plan should state when writes are paused and how records created since the backup will be protected. Reverting a commit does not resolve every failure.
PrestaShop provides a post-update checklist. Adapt it to the installation rather than treating it as a replacement for testing the custom checkout and integrations.
4. Choose representative cart scenarios
Start with a normal purchase, then add conditions that can change the outcome. For changes to the sales workflow, useful checks include:
- guest and signed-in purchases, including an important wholesale group;
- simple and variant products, correct price, quantity, availability and quantity changes;
- products without promotions, discounted products and vouchers, including valid combinations and exclusions;
- carts below, at and above free-delivery or other thresholds;
- domestic and international addresses, supported currencies and displayed taxes;
- returning to address, shipping or cart editing without losing the correct state;
- mobile and desktop devices and browsers actually used by customers.
Not every small change needs every combination. Select cases relevant to its scope and retain a mandatory basic purchase check. Record what was not tested too. A visible gap in acceptance is more useful than an undocumented claim that everything works.
5. Compare prices across listings, cart and order
Compare a product’s price in its category, on its detail page, in the cart and in the order summary. Include promotions, variants and customer-group conditions. If filtering or sorting is affected, check both directions and transitions between result pages.
Our PrestaShop price-sorting case study explains why promotional prices and index data needed separate verification. A correct product page did not establish the order of the entire listing. Likewise, a current-state check did not replace a later check at the scheduled start or end of a promotion.
6. Distinguish payment selection from payment confirmation
Reaching the payment provider verifies only part of the integration. In the agreed test mode, check successful, cancelled and failed payments and the response sent back to the store. Compare the amount, currency and order status. Check that a repeated notification does not repeat the same operation.
The test procedure depends on the provider and module. For example, the Revolut PrestaShop integration documentation distinguishes Sandbox and Live modes and identifies limitations of the payment methods available in each. Not every method can be exercised identically in a test environment. Make any untested Sandbox scope explicit in the production acceptance plan.
A real payment, refund or shipment is a separately agreed operation. Do not trigger one accidentally from a store copy. Production acceptance should identify the person responsible, the scenario and how the test order will be handled afterwards.
7. Shipping, multistore and integrations need their own checks
For delivery, check charges, availability, pickup points, cash on delivery and the data passed into shipment creation. In multistore, run the scenario for every affected shop and verify that another shop has not inherited unintended settings. Our guide to payments and shipping in PrestaShop multistore explores those dependencies.
When orders enter an ERP or an integration application, trace the next steps too. Saving an order in the store, importing it into an integration tool and creating it in a warehouse system are separate stages. Check identifiers, variants, quantities, amounts, delivery, statuses and error handling. Retrying a job should not create another order when the workflow is intended to process it once.
Our PrestaShop and WAPRO integration project illustrates related dependencies. Acceptance must extend beyond the frontend even when the visible change appears small. Check emails, sender settings and documents associated with the order as well.
8. Record a result someone else can reproduce
Each scenario needs conditions, an expected result, an actual result, a date and a status: passed, failed or not performed. Include reproduction steps and a screenshot without customer information for failures. Record the version or release identifier so the tested build is clear.
A useful note might read: ‘Shop A, guest, Polish address, PLN, cash-on-delivery pickup service. After selecting a point, only payment on receipt is available and the order amount matches. Sandbox label checked. No real shipment created.’ This distinguishes observations from operations outside the test scope.
For a bug fix, reproduce the original case first, then check the corrected result and neighbouring processes. Automating repeated checks helps, but does not remove the need to verify provider settings and behaviour important to the customer.
9. Deployment involves more than copying files
Plan the sequence: code release, any database migrations, module update, necessary index or cache rebuild, and restoration of scheduled tasks. Perform only the steps required by the change. Check production email, payment and shipping settings so test configuration is not left active.
After deployment, run a short acceptance check of the critical path and review logs and integrations. Agree an observation period and a way to report problems. State explicitly if no real payment or complete order was tested. Do not describe that stage as full confirmation of the sales workflow.
Need to organise testing and deployment? Explore Projectic technical support or describe the planned store change. We can agree backups, a test environment, scenarios and production acceptance, so deployment decisions rest on specific results.