The default Black Friday discussion normally starts with a percentage. Competitors offered twenty percent last year, trading wants something stronger this year and the final decision becomes a debate about whether the site should run twenty-five or thirty percent off, even though neither number says anything about product margin, stock, customer behaviour or the order the business actually wants to create.

A sitewide discount is easy to understand and easy to set up, but it applies the same logic to every product. It reduces the price of bestsellers that may have sold without help, puts more demand onto products with limited stock and can make low-margin orders actively unattractive once media, fulfilment, commission and returns are included.

I would begin with product economics. Map contribution margin, stock cover, return rate, repeat behaviour and full-price demand, then decide which products need an incentive and which should be protected. The best offer is not necessarily the biggest visible discount, it is the one that creates useful incremental demand without paying customers to buy what they were already going to buy.

Bundles can work well for beauty because they create a clear routine and move more than one product, while fashion brands may use selected-category offers, spend thresholds or value added through delivery and loyalty. The mechanic should fit the commercial problem. If the goal is clearing seasonal stock, a targeted markdown makes more sense than discounting newness. If the goal is acquiring customers into a replenishable range, use an entry product or routine that can lead to a second order.

Thresholds need modelling against the current basket distribution. A spend level just above the existing average may encourage customers to add something useful, while a threshold set too low gives the reward to orders that would already qualify and one set too high does nothing. Include the cost of gifts, free delivery, affiliate commission and expected returns rather than reporting only the larger checkout value.

The rules also need to be understandable. Complicated exclusions, different codes and changing prices create support work and abandoned baskets. If a product is excluded, say so before checkout. If the offer ends at a particular time, use one clear deadline across the website, email, paid media and social content.

Before launch, test real baskets using full-price products, sale items, bundles, gift cards, loyalty rewards and existing codes. Check how the offer behaves when stock runs out or a customer edits the basket. The most commercially intelligent strategy still fails if the discount does not apply consistently.

Report revenue with contribution margin, discount cost, new customer rate, product mix, return rate and the performance of customers after the event. A promotion that produces a large revenue number but attracts low-quality orders, strips margin and creates no repeat purchase should not be called a success because the dashboard looked busy for four days.

Choose the offer after understanding the economics. A round percentage is a mechanic, not a strategy.