Black Friday Discount Math: The Revenue Number That Lies to You

A record BFCM sales number can hide weak margin. Here's how to tell actual incremental growth from a dashboard celebration. Here's the practical operator's view, including the customer experience and the metric that matters.

The biggest number usually wins the meeting

Every November, someone announces that promotional revenue reached an all-time high. Everyone claps. Fair enough. But a total sales figure does not tell you how much profitable demand a campaign actually created. If discounts are doing backflips, you need another page in the deck.

A promotion can claim credit for a customer who already planned to purchase. It can pull a December purchase into November or lower the price paid by a customer who would have bought at full price. The attribution graph looks festive. The economics have a different story.

Start with contribution, not screenshots

Start with net sales after discounts and refunds. Subtract variable product costs, payment fees, fulfillment, subsidized shipping and any other costs that move with the order. Keep the incentive visible instead of burying it below a revenue headline.

Simple illustration, not an industry benchmark: two customers each generate $100 of gross merchandise value. One purchase requires a $25 price discount plus $10 in additional variable costs. Their reported gross sales might look similar, but the contribution each leaves behind is not.

A holdout beats heroic attribution

Where practical, randomly hold out eligible customers from an incremental offer and compare contribution margin per eligible person over an appropriate window. Track repeat purchases and refunds as well as same-day orders. The right window depends on the product's normal buying cycle.

Segmentation matters. Giving a large incentive to a customer who bought yesterday is a different investment than presenting a relevant winback to someone absent for a year. Evaluate first-time buyers, active customers, and lapsing subscribers separately.

Don't acquire the same discount customer twice

Measure holiday acquisition cohorts by their first promotional depth and whether they return at a healthier price point. If a second order requires another giant discount, include that in the initial acquisition economics. Cheap first orders can be surprisingly expensive introductions.

Great onboarding, shipping clarity, useful education and flexible replenishment often make a better second-order strategy than another loud offer. More discount is not the same thing as more perceived value.

A two-page executive scorecard

Page one: demand, orders, revenue, acquisition cost and offer mix. Page two: margin, incrementality estimate, refunds, fulfillment issues and first-to-second purchase conversion over 60–90 days. A complete view makes it possible to celebrate strong performance without pretending all revenue is equally valuable.

Nobody wants to ruin Black Friday. We just want to know whether it made the business stronger. Related: https://www.johntbrown.org/post/attributed-revenue-vs-incremental-revenue

How I would test it

Write down the customer problem, eligible cohort, control experience, treatment, success metric and economic guardrail before touching the automation. The first useful test is often small: one portal control, one educational message, one clarified promise, or one change to a pre-renewal reminder. Make the change reversible and check mobile QA before launch.

Measure successful fulfilled orders and retained contribution margin across a window appropriate for the purchase cycle. Use clicks, opens and portal entries as diagnostics rather than a substitute for customer outcomes. Review refund volume, care complaints and unexpected opt-outs to make sure a superficial conversion gain is not disguising a worse experience.

Assign one owner to report the results, including what did not work. If the effect is promising, expand methodically; if it is inconclusive, be comfortable admitting that. Great lifecycle strategy is a sequence of sound decisions, not a competition to produce the most elaborate dashboard.