What a discount actually costs

Twenty percent off is not twenty percent. If you sell something for $50 that costs you $15 to deliver, you keep $35 on each sale. Take twenty percent off the price and you keep $25. The customer saved a fifth. You gave up nearly a third of the profit on every order, including every order that would have happened anyway at full price.

That last clause is the expensive part, and it is the one nobody puts in the plan. To end the promotion with the same money in the bank, you do not need twenty percent more sales. You need forty percent more, because $25 has to be earned twenty-eight times to match what $35 earns in twenty. A discount that lifts orders by fifteen percent and feels like a great week is a week where you worked harder and earned less.

The arithmetic nobody runs before, everybody runs after

Do it once, on paper, before the offer exists. Take your price, subtract what each sale actually costs you to fulfil, and you have the number the discount comes out of. Divide the old margin by the new one and you have the lift you need just to stand still. At a seventy percent margin, twenty percent off needs forty percent more volume. At a forty percent margin, that same twenty percent off needs double. The thinner your margin, the more violent the discount, which is exactly backwards from how most people reach for one — thin margins feel like a reason to chase volume.

Now do the honest version. Of the people who bought during the sale, how many were waiting for a reason and how many would have paid full price on Thursday? You will not get a clean answer, but you can get a useful one: compare the sale week to a normal week, and treat the difference as the only sales the discount actually created. Everything else is a rebate you handed to people who had already decided.

The cost that outlives the week

The money is recoverable. The lesson is not. Run a sale in March and a sale in June and a sale in September, and by October a meaningful part of your audience has learned something true about you: waiting works. They are not being difficult. You taught them, clearly and repeatedly, that patience is worth a fifth of the price.

You see it first as a dead full-price week right before the promotion you have not announced yet, which is a strange and slightly eerie thing to watch. You see it next in how the sales themselves perform — each one a little flatter than the last, because you are increasingly selling to the same people who were always going to convert, only later and for less.

a discount is not a price change, it is a lesson your buyers remember

When it is genuinely the right tool

Three cases hold up. The first is buying a first purchase from people who have never heard of you — a discount aimed strictly at strangers is acquisition spending with a different name on it, and it should be judged against what you pay for a customer anywhere else, not against your margin chart. The second is time-bound inventory, in the broad sense: a cohort that starts on a date, seats in a session, anything whose value goes to zero if it goes unsold. The third is a deadline that is real. Not a countdown you invented, but an actual change — a price going up, a version retiring, a window closing. Readers can tell the difference, and the invented kind spends credibility you will want later.

What all three share is a boundary. The offer applies to a defined group, for a defined reason, ending on a defined day, and the reason is legible to the person reading it. An unbounded, unexplained "20% off this week" is the version that trains the calendar.

Things that move the decision without moving the price

Before cutting, try the cheaper levers. Add something instead of subtracting: an extra month, a bonus resource, faster onboarding — things with a real perceived value and a small marginal cost. Change the terms rather than the total: monthly instead of annual, a smaller first commitment, a guarantee that removes the risk of being wrong. Or fix the thing the discount is quietly compensating for, because a price objection is frequently a clarity objection wearing better clothes. When someone does not understand what they get, everything looks expensive.

And if you do run one, write the end date before you write the copy. Offers that drift past their own deadline are how a promotion becomes a price.

Judge it on four weeks, not on one

A promotion pulls demand forward, so the week itself always looks good and the week after always looks bad, and both are the same event. Measure the fortnight before and the fortnight after against a comparable stretch with nothing running. If the whole month is up, the offer created demand. If only the middle is up, it moved demand around and charged you for the privilege.

Measure it against money, not against activity — the record that only changes when someone actually pays. This is a place where the number you watch has to be the boring one: orders and revenue over a month, not clicks or redemptions during the sale. We pull revenue from billing and analytics tools you connect and show the period around a promotion rather than the promotion itself, and when there is not enough there to say anything yet, we would rather show you an honest empty panel than a flattering slice. A number that only looks good inside the window you chose is not a result.

So here is the whole thing in one line: a discount is the most expensive marketing you can run, because you pay for it out of the sales you already had — spend it on strangers, bound it with a real reason and a real date, and measure it over the month rather than the week.