Traffic spikes during a big sale expose problems that do not show up under normal day-to-day load, which is exactly why it is worth testing deliberately beforehand rather than finding out live.
Checking that hosting and server resources can actually handle the expected concurrent traffic is the first and most important step. A site that works fine for a hundred simultaneous visitors can fall over entirely at a thousand. Load testing ahead of a known high-traffic event is far cheaper than discovering the limit during the sale itself.
Payment gateway reliability under load is worth verifying separately, since a checkout failure at the final step is the most expensive kind of technical issue, the customer was ready to buy. Inventory sync accuracy matters too, since overselling a limited-stock item creates a customer service problem after the fact that is harder to fix than preventing it upfront.
None of this needs to be complicated, but it does need to happen before the sale starts, not during it.