Why the blanket discount survives
The storewide sale endures because it is easy, not because it works. It requires no segmentation, no model, no data infrastructure and no uncomfortable conversation with anyone about who deserves what. It can be decided in a meeting and live in the window by Friday. Best of all, it reliably produces a number that looks like success: revenue up, units up, footfall up, a graph pointing the right way just in time for the Monday review. Every one of those numbers is real. None of them answers the only question that matters, which is how many of those buyers would have paid full price anyway. That number is never in the report, because measuring it requires deliberately not discounting to a comparable group — and nobody wants to be the person who suggests leaving money on the table to find out. So the sale gets repeated, the graph goes up again, and the organisation slowly convinces itself that a reflex is a strategy. The uncomfortable truth is that a blanket discount is not a decision about pricing at all. It is a decision to stop thinking about pricing, purchased with margin.
What a discount actually costs you
Every price cut is spent twice, and the second payment is the one that ruins you. The first cost is visible and accepted: margin surrendered on the units you sold. It appears in the P&L, it was forecast, and everyone has made their peace with it. The second cost is invisible and cumulative — you have just taught your best customers to wait. Shoppers are excellent pattern matchers. A sale every six weeks is not a promotion; it is a schedule, and once it is learned, full-price demand quietly migrates into the discount window. The sale stops creating incremental demand and starts harvesting demand you already had. This is the trap that makes discounting so hard to escape: the more reliably you run sales, the more of your baseline they absorb, and the worse full-price weeks look by comparison — which makes the next sale feel necessary. Each round is individually defensible and collectively fatal. Eventually the discount is no longer a lever you pull to change behaviour. It is a tax you pay on your own calendar, levied by customers who have simply learned your habits better than you have.
Targeting the margin, not the crowd
A discount is only worth giving to someone whose behaviour it changes. Stated plainly it sounds obvious; applied honestly it eliminates most of what retailers actually do. In any given sale, your buyers fall into two groups. The first was always going to buy — they had chosen, they were walking to the till, and the discount simply reduced what they handed over. For them, the offer bought nothing. It was a rebate on a decision already made. The second group was genuinely undecided: the hesitant, the lapsed, the price-sensitive, the shopper standing in front of something above their usual spend who needs one reason to commit. That group is where discounting earns its keep, and it is invariably far smaller than the group you gave the discount to. Behavioural data is what separates them, and the signals are not subtle once you can see them — dwell without purchase, repeat visits to the same fixture without a basket, a lapsed regular back in the store for the first time in months. None of this requires a perfect model. It requires only that you stop treating everyone in the building as though they were identical, which is precisely what a storewide sale does by definition.
Building a policy that compounds
Replace the calendar with rules. A rule has three parts and fails if any one is missing. First, a trigger tied to observed behaviour rather than a date: a dwell that ends without a purchase, a regular who has not returned in ninety days, a basket about to walk. Second, a guardrail — the floor beneath which the offer destroys more value than it creates, set once, in advance, by someone who is not under pressure to hit a weekly number. Third, and non-negotiably, a holdout: a comparable group who never receive the offer at all. Without the holdout you can prove only that people like cheaper things, which was never in dispute. The holdout is what converts "the sale went well" into "the sale caused eleven percent incremental units, and here is the group that proves it." Retailers resist this because it feels like deliberately forgoing revenue, and for a small slice it is. It is also the only mechanism that will ever tell you which of your promotions are working, which means it pays for itself the first time it kills one that never did.
- A discount given to someone already buying is a rebate, not a strategy.
- Blanket sales train your best customers to wait for the next one.
- Without a holdout group, you cannot prove a discount did anything at all.