DEV Community

Cover image for WooCommerce 9.8 11.3: What a Real Test Order Showed
Manh Nguyen
Manh Nguyen

Posted on Originally published at updatedrill.com

WooCommerce 9.8 11.3: What a Real Test Order Showed

Checkout kept working after a jump from WooCommerce 9.8.0 to 11.3.0, and the order total and shipping stayed exactly the same. What did change was the layout: 12.1% of the checkout page and 6.8% of the empty cart page looked different afterwards.

We tested this jump on purpose. Many stores sit on an old version for months, then update everything in one evening. That is the riskiest kind of update, and the one people are least likely to test by hand.

The test setup

The test ran on a fresh sandbox store in Docker, not a client site. It is deliberately plain, so any change comes from the update itself.

Setting Value
WooCommerce 9.8.0 → 11.3.0
Theme Storefront 4.6.0 → 4.6.2
Checkout Block checkout (the default for new stores)
Products 1 simple product, 150,000 VND
Shipping Flat rate, 30,000 VND
Payment Cash on delivery
Currency Vietnamese dong (VND), no decimals

The tool took a snapshot, loaded the home, shop, product and cart pages, and placed a real test order. It then applied each update on its own, repeated every step, and compared the two runs: page status, order total, shipping, screenshots, browser console errors and new lines in debug.log.

Results

The run ended as a WARN, not a FAIL: nothing broke, but two pages looked noticeably different.

Check Before (9.8.0) After (11.3.0) Verdict
Test order placed Yes Yes ✅ PASS
Order total 180,000 VND 180,000 VND ✅ PASS
Shipping 30,000 VND 30,000 VND ✅ PASS
New PHP fatal errors — None ✅ PASS
Empty cart page pixels changed — 6.8% ⚠️ WARN
Checkout page pixels changed — 12.1% ⚠️ WARN

All five pages returned HTTP 200 before and after. There were no new JavaScript console errors and no new PHP log entries. The whole run took 63 seconds.

The warning threshold is 2% of pixels. Layout changes are a warning rather than a failure, because they are often intended. A person should look at them before the update reaches a live store.

As a control, a run that updated only Storefront passed with no warnings.

What changed on cart and checkout

On this Storefront setup, with its default sidebar, the checkout went from two columns to one. The page grew from 2,077 to 2,273 pixels tall. Three things moved:

  • Order summary. Before, it sat in a narrow column to the right of the form. After, a collapsed bar at the top shows the total, and the full summary moved below the payment options.
  • Place Order. Before, it was a button at the end of the form, next to a "Return to Cart" link. After, it is a full-width button at the very bottom, and the "Return to Cart" link no longer shows.
  • Terms notice. "By proceeding with your purchase…" moved from just above Place Order to just below the payment options.

Checkout page before (WooCommerce 9.8.0), after (11.3.0), and the changed pixels highlighted in red
Checkout before (9.8.0), after (11.3.0), and the changed pixels in red.

The cart screenshots show the empty cart, because the test opens /cart/ in a fresh browser. WooCommerce 11 dropped the large sad-face icon and shows a bigger "Your cart is currently empty!" heading instead. The "New in store" products moved up, and the page got 84 pixels shorter.

Empty cart page before (9.8.0), after (11.3.0), and the changed pixels highlighted in red
Empty cart page before (9.8.0), after (11.3.0), and the changed pixels in red.

In the diff images, red marks changed pixels. The solid red band at the bottom is the difference in page height. Most of the 12.1% on checkout is content that moved, not content that changed.

None of this stopped the order from going through. On a store with custom CSS, a checkout extension or a page builder, the same shift could hide a field or push the Place Order button below the fold. That is why a screenshot comparison sits beside the test order.

If you automate checkout tests: wait for the recalculation

Our first automated run reported "Checkout broke after the update". Checkout was fine; the test clicked too early.

After an address is entered, block checkout recalculates the cart in the background. On WooCommerce 11 this sometimes happens in two passes, and the first shows nothing on screen. A Place Order click during that window is silently ignored.

What fixed it:

  1. Wait until WooCommerce's background requests have finished before reading totals or clicking Place Order.
  2. After clicking, confirm that an order request actually went out.
  3. If none did, click again, up to 3 times. Never click again once a request has gone out, so you can't create duplicate orders.

If you test with Playwright, Cypress or similar, a fixed sleep is not enough. Wait for the network, then verify the request.

Updating after a long gap: a checklist

  1. Copy the live store to staging, and block outgoing email there so test orders never reach customers.
  2. Place a test order on staging before updating, and note the total and shipping.
  3. Apply updates one at a time, so a problem points at one plugin.
  4. Run WooCommerce's database update after updating the plugin.
  5. Place the same test order again and compare the total and shipping to the cent.
  6. Compare cart and checkout screenshots, especially with custom CSS or checkout plugins.
  7. Read the new lines in debug.log, not the whole file.
  8. Only then update the live store, with a backup you know how to restore.

To check that a test like this actually catches problems, we broke the sandbox on purpose. A small plugin raised shipping from 30,000 to 45,000 VND once WooCommerce reached version 10. The run failed with "Shipping changed" and restored the snapshot.

About UpdateDrill

UpdateDrill runs every step of that checklist for you. It clones the store to staging, places the test orders, compares the results and rolls staging back on a failure. You get a pass, warn or fail report you can forward to a client.

It's in early access. If you look after WooCommerce stores, request a free check and we'll run one update rehearsal on a store of your choice. We'll repeat this test for each new WooCommerce release.


Originally published at updatedrill.com.

Top comments (2)

Collapse
 
alhiane profile image
Alhiane Lahcen •

Useful write-up. The silently ignored Place Order click is exactly the kind of thing that makes people stop trusting automated checkout tests. A few ideas from running similar before/after checks:

  • Pixel diffs and field diffs catch different things. A missing checkout field (a VAT number or delivery-date field added by another plugin, say) can change only a few percent of pixels, under a 2% threshold if it sits low on the page. Saving the name attributes of every input inside the checkout form before and after, and diffing the two lists, flags it exactly.
  • To confirm an order really went through, I'd check the store side as well as the network request: compare the newest order ID before and after the click (wp wc shop_order list --user=1 --per_page=1 --field=id). That also makes the retry rule safer: only retry if no new order exists.
  • After the WooCommerce database update, look at WooCommerce > Status > Scheduled Actions for pending or failed actions. A big version jump can queue migration jobs that are still running when the test order is placed.

Will you run the same check on classic checkout (the shortcode) next? That's where I'd expect custom CSS and checkout-field plugins to differ most.

Collapse
 
manh_nguyen_0805f1ab35f00 profile image
Manh Nguyen •

Thanks, these are really useful. The field-name diff is a great idea, a missing VAT field could easily hide under the 2% threshold. Adding it. Same for checking the newest order ID, with one catch: block checkout creates a checkout-draft order on every visit, so drafts have to be skipped. Hadn’t thought about scheduled actions after the DB update either. And yes, classic checkout is next.