How to choose a Black Friday discount without guessing

• A discount comes out of your margin, not your price. A 20% discount on a 50% margin takes away 40% of what each order earns.
• Work out the break-even: how many more orders the discount needs just to earn what you earn today. For a 20% discount on a 50% margin, that's two-thirds more orders.
• Compare % off, $ off and a gift on what they cost you per order, not on how big they sound.
• Don't A/B test discount depth at small-store traffic. You can't detect the differences that matter, and shoppers who get the smaller discount will notice.
• Compare with last year, with context: the discount, the traffic, the order count and the margin, written down side by side.

Every order earns a contribution: the price minus the cost of the product. Your gross margin is that contribution as a share of the price. A percentage discount takes its share off the price but nothing off the product cost, so it comes entirely out of the contribution.

The number of orders you need to earn the same total as before is:

break-even increase in orders = discount ÷ (margin − discount)

This is the break-even sales change from pricing analysis (Nagle and Müller, The Strategy and Tactics of Pricing), and it's what our glossary entries on break-even analysis and contribution margin describe.

Plain arithmetic, from the formula above. Margin here is gross margin on the full price; payment fees and shipping you pay for make the real break-even a little higher.

Read the table as a question: "Will 30% off really bring me two and a half times as many orders as I'd get without it?" For a store with a 50% margin, that's what 30% off needs to earn the same money. Some stores clear that bar during Black Friday. Many don't, and sell more while earning less.

There are good reasons to accept a lower profit per order in a sale: clearing old stock, winning first-time customers who buy again, or making room for new products. Those are fine reasons, as long as you know the cost going in and write the reason down.

The form of the offer changes what it costs you, especially when order sizes vary.

What the example shows:

• % off scales with the basket. A shopper buying $150 gets $30 off. That's generous on large orders, which may already have come without a discount.
• $ off is a fixed cost per order and costs you a smaller share on bigger baskets. With a minimum ("$10 off orders over $50") it can also nudge baskets up.
• A gift costs you its product cost, not its price. A $20 gift that costs you $8 can feel more generous than $10 off while costing less. Check the extra shipping weight, and pick a gift you have plenty of.
• Tiers ("spend $75 save 10%, spend $120 save 15%") work like a threshold. They cost little on small orders and reward bigger ones. Run each tier through the formula.

Testing 20% off against 30% off for different shoppers looks like the obvious way to settle it. Two things stop most small stores.

The traffic. Suppose you want to know whether a deeper offer brings 25% more orders. At a 2% conversion rate, a two-version test needs 13,809 visitors per version to detect that at 95% confidence and 80% power (Benchmyrk's sample-size formula). At 2,000 visitors a week, that's about 14 weeks, and Black Friday is six weeks away. And whether the deeper discount pays isn't a conversion-rate question: it's a profit-per-visitor question, which needs even more visitors because order values vary.

The shoppers. Deals are shared and screenshotted. A shopper who gets 20% off and sees a friend's 30% off will ask for the difference, and they'll be right to feel treated unfairly. In a sale week, that's a support problem and a reputation problem.

What you can test, if you have the traffic, is the format at the same cost, before the sale: 20% off against $12 off on a $60 average order, for instance. Same cost on the average basket, different message. Start by 25 October. Most small stores are better off choosing with the arithmetic above.

If you ran a Black Friday sale last year, last year is your best comparison, as long as you write down what was different. Use Shopify's sales reports and fill in this table for last year's sale weekend and a normal weekend before it:

Then ask: did the sale bring enough extra orders to cover the discount, by the break-even formula?

This comparison is rough. Your traffic, your email list and your products changed in a year, so it's a guide, not a test result. Part 8 of this series explains why before-and-after comparisons mislead in a sale week.

• Choosing a depth because others in your niche use it. Their margins are not yours.
• Stacking discounts by accident. Shopify discounts only combine when you allow it, by class (product, order, shipping). Check your sale discount against free shipping, email codes and any loyalty discount, and place a test order.
• Forgetting the end date. A discount without an end date doesn't expire.
• Judging the sale on revenue alone. Revenue rewards the deepest discount. Look at what each order earned after the discount.
• Changing the offer during the sale. Shoppers who bought on Friday and see a deeper offer on Monday will ask for a refund of the difference.

• [ ] Gross margin known (or a conservative estimate)
• [ ] Break-even worked out for two or three offers
• [ ] Offer chosen and its reason written down
• [ ] Discount created with start and end dates
• [ ] Combinations checked with a test order
• [ ] Last year's sale numbers in the table above, if you have them

Choosing offers for client stores? The agency track on the BFCM hub covers offer tests for clients: sign-off, margins and discount conflicts. For the general BFCM plan, see what to test before Black Friday, and for the trade-off between price and discount tests, test prices or discounts first?