How to know your checkout still works after a deployment

The short answer

Right after the deploy, make one test purchase in a private window: a real product, add to cart, checkout, an address, a shipping method, up to the payment step. If the store allows it, place a real order and cancel it. Then keep a scheduled check running, because some failures appear only hours later, when caches refill, cron jobs run or traffic arrives from another country.

In our checkout failure data from 89 stores, a third of incidents (32%) started between 22:00 and 03:00 Athens time, the hours when deploys, cron jobs, indexers and imports usually run.

The five-minute test purchase

1. Private window. No admin session, no old cookies, no cached cart.
2. A real product. One that is in stock and has a price; with options, pick them.
3. Cart. The item is there, with the price the product page showed.
4. Checkout. The page opens and the address form accepts a normal address.
5. Shipping. At least one method appears for that address and can be selected.
6. Payment. The methods appear and the card form loads. On a store that allows test orders, place one and cancel it in the admin.

Keep the browser console open while you do it. A red error at the moment a step fails is the fastest clue you will get.

Why a test right after the deploy is not enough

Some failures need time to show up. The full page cache serves the old pages until it expires, and then the new ones break. A cron job or indexer runs at night and changes prices or stock. A third-party script loads a new version without any deploy on your side. A rate limit on a payment or shipping API only trips under real traffic.

That is why the check has to repeat. Run the same test purchase on a schedule, every 15 or 30 minutes, so a failure that starts at 02:00 is found at 02:15 and not when the first customer writes in the morning.

Automate the smoke test in your pipeline

If you deploy through CI, a short browser test after each deploy stops most bad releases before customers see them. Playwright is a common choice. This sketch walks a WooCommerce classic checkout to the payment step:

// checkout.smoke.spec.ts — run after each deploy: npx playwright test
import { test, expect } from '@playwright/test';

const STORE = process.env.STORE_URL!;           // e.g. https://example.com
const PRODUCT = process.env.PRODUCT_PATH!;      // e.g. /product/blue-mug/

test('a shopper can reach the payment step', async ({ page }) => {
  await page.goto(STORE + PRODUCT);
  await page.click('button.single_add_to_cart_button');
  await page.goto(STORE + '/checkout/');
  await page.fill('#billing_first_name', 'Test');
  await page.fill('#billing_last_name', 'Shopper');
  await page.fill('#billing_address_1', '1 Test Street');
  await page.fill('#billing_city', 'Athens');
  await page.fill('#billing_postcode', '10558');
  await page.fill('#billing_phone', '2100000000');
  await page.fill('#billing_email', 'test@example.com');
  await expect(page.locator('#shipping_method li').first()).toBeVisible();
  await expect(page.locator('.wc_payment_method').first()).toBeVisible();
});

Scripts like this need upkeep: every theme change, checkout plugin or new required field can break the selectors, and then the test fails while the store works. Keep it short and fix it the same day it breaks, or people stop trusting it.

When the test fails: roll back first

If the checkout broke with the deploy, roll the deploy back, then investigate. Every hour spent debugging on production is an hour without orders. Once it is back, compare the two releases on staging.

The most common causes have their own guides: five ways a Magento deploy breaks checkout (static content, caches, module updates, configuration drift) and, for WooCommerce, checkout totals that stop updating. If payment is what fails, read is the payment gateway down, or is it my store?.

Have the check run for you

MageSentinel runs this test purchase on a schedule, every 30 minutes or faster depending on the plan, in a real browser that knows Magento and WooCommerce checkouts, with no script to maintain. When a step fails you get one alert with the step, a failure code, a screenshot, the console and the network log. After a deploy, press Run now on the monitor in the dashboard to check straight away instead of waiting for the next scheduled run.

See everything MageSentinel checks, or try the free checkout audit on your store now.