The weeks before Black Friday create pressure to finish every improvement that has been discussed during the year. A new navigation, redesigned product page, additional apps, new discount logic and a different checkout message all get pushed towards the same deadline because the team wants the site to be at its best, but changing several important parts of the journey at once makes it harder to test properly and much harder to understand what caused a problem.

Peak is not a useful deadline for work that is not ready. If a change affects product discovery, pricing, basket rules or checkout, it needs enough time for testing, data review and customer service feedback before the highest-traffic period begins. A feature that looks complete in staging can still behave differently with real products, discount combinations, devices and customer accounts.

Create a release freeze for high-risk work and define what qualifies as an emergency change. Content, merchandising and planned promotional updates can continue, but structural changes should require a clear commercial reason, an owner and a rollback plan.

The same discipline applies to apps and scripts. Every new popup, tracking tag, personalisation tool and promotional widget adds weight and another possible point of failure. Review what is running before peak, remove duplicate or unused tools where safe and make sure essential tracking is tested after any change.

Discount setup should be completed early enough to test with real basket combinations. Check sale products, gift cards, bundles, loyalty, free delivery, subscriptions and any creator or affiliate codes that may still be active. Customers should not be able to stack offers that were meant to be separate, but genuine promised discounts also need to work without support intervention.

Ownership needs to be obvious once the event starts. Decide who can change the homepage, pause media, alter merchandising, update delivery messages and respond to a technical incident. If every decision needs a meeting, the team will move too slowly. If everybody can change everything, the site will become inconsistent and nobody will know which version is correct.

Keep a simple log of changes during peak, including what changed, when, why and who approved it. This makes it easier to connect performance shifts with site activity and gives the team a record when reviewing the event afterwards.

The aim is not to stop improvement. It is to protect the parts of the customer journey already working while the business is under the most pressure. A stable site with clear products, accurate offers and reliable checkout will normally outperform a more ambitious experience that has not been tested properly.

Do the risky work early, freeze what needs protecting and keep November changes small enough to understand.