Businesses discount to solve a specific problem: an empty appointment, a first visit that has not happened, stock that will soon lose value, a regular customer they want to keep, or an offer nobody has noticed. A higher sales count alone does not tell you which problem was solved—or whether the discount earned more than it cost.

The useful question before any promotion is: What would have happened without it?

What job is the discount doing?

1. Fill capacity that expires. A restaurant table at 3pm or a salon chair on Tuesday morning cannot be sold tomorrow. Measure contribution from the targeted hours, then check whether full-price customers simply moved into the cheaper slot.

2. Win a first visit. A trial price can reduce the risk of trying an unfamiliar venue. Measure visits that would not otherwise have happened and whether those customers return without the introductory price. “First time in the CRM” is only a proxy for an incremental visit.

3. Clear ageing stock. A retailer can recover cash before seasonal goods need a deeper markdown or write-off. Compare contribution from the sale with the realistic next-best option—not the original ticket price.

4. Reward regulars. A benefit can give existing customers a reason to stay. Compare their visit frequency and contribution with their own history or a comparable group, after the reward cost.

5. Get noticed. A discount can make a business more visible in a crowded category. Count the extra visits attributable to that exposure, not just offer views or redemptions. An offer that regulars use but nobody new discovers is a different investment from acquisition.

These jobs overlap, but they should not share one scorecard. A first-visit offer available to every regular may quietly become a loyalty expense. A weekday offer that also applies on a sold-out Friday is paying to fill seats that were already full.

Why isn't a sales spike proof?

Some customers would have bought at full price. Others move a purchase from a different day. Others switch from one of your full-price products to the discounted one. Only some demand is genuinely additional.

In a 2004 study of four grocery datasets, researchers found that, on average, roughly a third of a promotion's sales lift came from each of three sources: switching between brands, borrowing from other periods, and expanding the category. That is not a forecast for a Dubai café or salon. It is a reason to separate “more sales during the offer” from “more business because of the offer.”

Did this Dubai salon's morning offer pay?

Consider an illustrative, merchant-funded 30% discount offered Monday to Thursday mornings on a mani-pedi normally priced at AED 210 including VAT. At the UAE's 5% VAT rate, the salon receives AED 200 before VAT at full price and AED 140 before VAT on the offer. Suppose products and per-service pay cost AED 40 either way. Contribution is therefore AED 160 at full price and AED 100 on offer.

Morning bookings rose 150%, but weekly contribution rose only AED 560. If booking histories show that 24 morning customers would have visited anyway, 10 moved from full-price slots, and 26 visits were genuinely additional, the arithmetic is straightforward: the first 24 cost AED 1,440 in markdowns; the 10 who moved cost AED 600; the 26 extra visits add AED 2,600. Net gain: AED 560. All customer movements and costs here are assumptions, not measured UAE benchmarks.

The result is positive, but narrow. Change the share of visits that were genuinely additional and the conclusion changes quickly. Future full-price repeat visits could improve it; future discounted repeats could weaken it.

What should a business do before launching a discount?

Write down the job and its success measure first. Limit the offer to the customers, products or hours that fit that job. Then compare total contribution during the offer with comparable hours, weeks or locations without it. Redemptions show use; they do not, by themselves, show incremental profit.

Discounts are not only an F&B tool. In DiscGo's September 2026 index, 61% of UAE venues with a live offer were outside dining. That is a count of indexed venues, not proof those offers brought customers in. The same standard applies everywhere: name the job, measure the extra contribution, and stop when the job is no longer being done.