E-commerce testing before Black Friday: what to check, and what to stop changing

E-commerce testing before Black Friday: what to check, and what to stop changing

In short: E-commerce testing before Black Friday is about the flows that earn money on the day. Think search, cart, discount codes, checkout and payment, plus the systems behind them. Load test the whole chain, not just the server. Then freeze changes, and let an automated regression set and a smoke test guard every exception you still allow.

Black Friday is on 27 November this year, and Cyber Monday follows on the 30th. With that little time left, the most useful question isn’t what you can still build. It’s what can still break, and how fast you’d know. E-commerce testing before Black Friday is mostly about that second part. We’d stop adding features now and spend the time on the checkout. A shop that’s a bit plain but works all day beats a shiny one that falls over at nine in the morning.

What should you still test before Black Friday?

Test the flows that earn money on the day, from first click to confirmation email. That’s search, the product page, the cart, discount codes, checkout, payment and the order confirmation. Then test what sits behind them: stock levels, prices, the payment provider and delivery options. Include the link to your ERP, the system that handles orders and stock in the back office. Most teams test the pages and forget the chain. The customer doesn’t care which part failed. They only see that their order didn’t go through. So write down your five or six money flows and check each one end to end, on mobile as well as desktop. Pay extra attention to anything that changes for the sale itself: new discount rules, bundles, free shipping above an amount, a countdown banner. Those are new code paths that have never seen real traffic. Our article on payment testing for webshops goes deeper into the checkout part.

One pattern we see at e-commerce clients is worth checking right now. New integrations get tested once, when they go live, and then never make it into the regression tests. If you added a new payment method or delivery partner this year, make sure it’s in your checks before the peak.

Why does a webshop break on a peak day?

A webshop usually breaks at its weakest link, and that’s rarely the web server. It’s the payment provider that slows down or the stock check that can’t keep up. Or a marketing script that blocks the page. Under normal traffic you never notice these links. Under ten times the traffic, the slowest one sets the pace for everything. And when one part slows down, others start waiting for it, retrying and piling up. Google’s engineers call that a cascading failure: one slow part drags the rest down with it.

The other big cause is change. A small fix pushed the week before Black Friday, a new tracking tag from marketing, a price import that runs at the wrong moment. Each one is harmless on its own. Together, at the worst moment, they’re how a good shop has a bad day.

How do you load test a webshop?

A load test simulates many shoppers at once, so you see where the shop slows down before real customers do. Use realistic journeys, not just the homepage: people who search, browse, add to cart, use a code and pay. Run it on an environment that looks like production, and step up the load until something gives. That point tells you how much headroom you have. A stress test goes further, past the expected peak, to see how the shop fails and whether it recovers. Grafana’s k6 guide explains the different types of load test well.

Two things matter most. First, include the external services, or agree with your payment provider how to test against them. A load test that skips payment tells you the part you were least worried about. Second, measure what the shopper feels, like how long checkout takes, and not only server numbers. The Web Vitals from Google are a good start for page speed. Our Online Retail case shows why the chain matters. There, performance testing covered the systems around a new ERP, not just the front end.

A trade-off to name: if you’ve never run a load test, ten days is short to build a good one. Then test the money flows by hand and with your automated tests, and plan extra capacity with your hosting party. Put a proper load test on the list for next year.

What does a change freeze need to work?

A change freeze, a period in which nothing goes live unless it has to, only works with three things. A clear start date, a short way to approve exceptions, and a fast check after every exception. Most teams have the first two. The third is where it goes wrong. An urgent fix still has to go live, and then someone clicks through the shop for an hour and hopes for the best.

  • A date. Pick it now and tell marketing too. New scripts and banners are changes as well.
  • An exception rule. Who decides, and what counts as urgent? Keep it to one or two people.
  • An automated regression set. The tests that check whether what worked yesterday still works today, for every money flow, run after every exception.
  • A smoke test on production. A few quick checks right after each deploy: can you search, add to cart and reach payment?

If your regression set isn’t reliable, the freeze turns into a gamble. Tests that fail at random get ignored, and real problems slip through with them. We also see tests fail because test data isn’t managed. A product is out of stock in the test environment, or a code was already used. Fix that before the freeze, not during it.

What do you watch on the day itself?

Watch the signals that show money coming in, not just whether the site is up. Orders per minute compared with the hour before, payment errors, the time a checkout takes, and whether stock updates keep up. A shop can be “up” while nobody can pay. Agree beforehand who watches, who can decide on a rollback, and how you reach your payment provider and hosting party fast. Write the rollback steps down. In the middle of a busy morning, nobody wants to work them out from memory.

Keep one person free to use the shop like a customer does, on a phone, a few times during the day. Automated checks tell you a flow works. A person notices that the discount banner covers the pay button on a small screen.

What this means for your team

A peak day shows which parts of your testing are reliable and which only exist on paper. The work that gets left behind is the same every year. Someone has to keep the regression tests up to date, add the new integrations, and keep test data under control. When that has no owner, the freeze rests on hope.

That’s the ongoing work we do with Test Automation as a Service, working alongside your team. For teams that still run their regression tests by hand, we cut that manual work by up to 70%. Your tests run every night, and we keep them reliable. You get a stable test suite your team can trust, one that doesn’t depend on a single person or add to your team’s workload. See our e-commerce testing page for how we look at the whole chain.

A first step for this week: list your five money flows. Next to each one, write how you’d know within five minutes that it broke. Every blank space is your Black Friday risk.

Regression tests you can trust on your busiest day.

See how TaaS worksTalk to our team

Frequently asked questions

When should you start a code freeze before Black Friday?

Most webshops start one to two weeks before, and keep it until after Cyber Monday. The exact date matters less than sticking to it, and having a fast test for the exceptions you still allow.

Is load testing on production a good idea?

Only with care, outside busy hours, and with your payment provider and hosting party informed. A production-like test environment is safer. Production does show the real setup, including the services you forgot about.

What if we can’t load test in time?

Test the money flows end to end, by hand and automated, and ask your hosting party about extra capacity. Prepare monitoring and a rollback plan. Then plan a real load test well before the next peak.

Does Black Friday testing only matter for big webshops?

No. Smaller shops often feel a peak more, because one slow plugin or payment link can stop every order. The checks are the same; the scale is smaller.