Black Friday falls on 27 November 2026, which leaves five weeks. That is enough time to prepare a store properly and not enough to start in the last fortnight, which is when most shops begin and why most of them spend the weekend firefighting instead of selling.
Contents
What to do in each of the five weeks
Work backwards from the date, because the order matters more than the list. Anything that touches prices has to be settled first, anything that touches infrastructure second, and anything that touches copy last.
Five weeks out, settle the offer and the legal side. Decide what is discounted, by how much, and confirm your price history is being recorded, because the lowest-price rule applies to the whole promotion and cannot be fixed retrospectively.
Four weeks out, test the store under load and audit the plugins. Both are infrastructure questions, both take time to fix, and neither can be done on the day.
Three weeks out, set up the discounts themselves and schedule them, so nothing depends on somebody being at a keyboard at midnight. Two weeks out, prepare checkout and fulfilment, which means the fields customers fill in and the labels you print afterwards.
The last week is for copy, emails and the things you will want to change twice. Leave it empty of anything structural.
The three things that go wrong
Prices announced before the history exists. A store that starts recording its price history the week it launches the promotion has no thirty-day history to show, and the discount is either unlawful or has to be presented against a reference price that makes it look smaller than it is.
A plugin stack nobody checked. The season is when stores discover that a plugin they stopped paying for is two years behind, and the discovery usually arrives as a checkout that fails for some customers and not others.
And manual promotions. A sale price typed in by hand on the Thursday is a sale price somebody has to remove by hand on the Monday, and the one that gets forgotten is always the highest-margin product in the catalogue.
What is worth skipping
A redesign. Whatever is wrong with the shop in October will still be wrong in December, and the week before a peak is the worst possible moment to change a layout that currently converts.
New plugins you have not run before. Anything installed in the last fortnight has not been through a full order cycle on your data, and Black Friday is not the week to find out what it does under load. If you do need something new, install it now and run it for three weeks first, ideally on a staging copy of the store.
Also worth skipping is the all-day countdown banner. It costs page speed, it annoys the customers who were going to buy anyway, and the stores that measure it rarely keep it the following year.
Most of the five weeks above comes down to four or five plugins doing their part on schedule. The All Plugins Bundle covers pricing, checkout, returns and analytics on one licence, which is usually cheaper than buying them separately in the week they turn out to be needed.
Summary
Five weeks before 27 November is the right moment to start, and the order of the work matters more than the list. Week five settles the offer and the legal position, including confirming that price history is being recorded, because the lowest-price rule covers the whole promotion and cannot be applied after the fact. Week four is infrastructure: a load test and a plugin audit, both of which take time to fix. Week three sets up and schedules the discounts so nothing depends on somebody being awake. Week two covers checkout and fulfilment. The final week is copy and email, with nothing structural in it.
Three failures recur. Announcing prices before the history exists, which leaves the discount either unlawful or presented against a reference price that understates it. An unaudited plugin stack, where a licence that lapsed two years ago surfaces as a checkout failing for some customers. And manual promotions, where a price typed in on Thursday has to be removed by hand on Monday and the one forgotten is always the highest-margin product. Three things are worth skipping outright: a redesign, any plugin you have not already run for three weeks, and the countdown banner.
Frequently asked questions
These three come up every October.
Is five weeks really necessary?
For the price history, yes, because thirty days is the reference period itself. The rest can be compressed, though compressing the load test is how stores find out about a problem on the day.
Should the sale start before Friday?
Many stores now run the whole week, and the data generally supports it. What matters more is that the start and end are scheduled rather than manual.
What should be ready on the day itself?
Nothing. If anything still needs doing on the Friday, it was scheduled too late, and the day should be spent watching orders rather than editing products.
