A discount changes the price and asks the buyer to judge a deal. A tiny saving can make an optional purchase look less worthwhile, even though its price fell. A deeper discount can win orders today while changing who buys or when they buy next.
Decide whether the saving is meaningful for the actual basket, whether the purchase is needed and what pricing history customers have learned. Keep a no-promotion comparison and follow the economics beyond the sale.
🔎 What research says
Small discounts sometimes backfired in controlled candy tasks, but the effect depended on quantity and purchase purpose. Price histories also changed remembered value. Separate paid buying from intentions and perceptions, and distinguish campaign response from later customer economics.
🗺️ In this guide
1. Test the money saved on the actual basket
📈 Recommendation
Keep a no-promotion control when testing a small discount. Calculate the money saved on the quantity people actually intend to buy.
For a €4.50 snack, 10% saves €0.45; five save €2.25. This is illustrative arithmetic. It does not show that buyers need five or that a larger pack will be profitable.
Compare paid orders and kept-order contribution against the undiscounted offer. Include the discounted margin on shoppers who would have bought anyway.
🎓 Findings
Cai, Bagchi and Gauri (2016) tested actual candy buying among 119 Chinese students. For one can, a 10% discount produced 19.23% buying versus 43.33% without a discount: a 24.10-percentage-point descriptive decline. This contrast was only marginal (p = .054). For five cans, buying was 26.67% versus 3.03% (p < .01). The package-by-discount interaction was clear (p < .01).
🧠 Why it works
A small saving can make an optional purchase look like an unimpressive deal. The same percentage can represent a larger saving on a larger basket.
✋ Limitations
One campus candy task cannot set a minimum discount. Cell counts and confidence intervals were not reported in the retained main text; do not turn the marginal one-can result into a certain decline.
2. Separate needed purchases from optional purchases
📈 Recommendation
Use observed shopping purpose to interpret a promotion test. A top-up for a planned task and an unplanned treat may respond differently.
For party supplies, distinguish visitors assembling a known shopping list from visitors browsing optional extras. Define those groups before evaluating the promotion.
Keep the product’s genuine use clear. Do not manufacture urgency or claim that a treat is essential because a laboratory scenario used that word.
🎓 Findings
In Study 4 of Cai and colleagues (2016), 124 analyzed students imagined buying candy for a party. With a 5% discount, mean buying intention was 2.67 versus 3.95 without one when the purchase was optional; when a friend’s preference made it more necessary, it was 5.53 versus 4.30. These are 0–10 intention scores, not buying rates. Both contrasts had p < .05; the interaction had p < .01.
🧠 Why it works
A necessary item can be attractive because it solves a task; an optional item may rely more on whether the deal itself feels worthwhile.
✋ Limitations
The scenario changed a friend’s candy preference. It does not classify all groceries as essential. Three incomplete responses were excluded; no confidence intervals were reported. Test paid behavior before changing a promotion policy.
3. Audit the promotion history buyers learn
📈 Recommendation
Evaluate today’s discount alongside the pricing pattern customers have been seeing. Equal average prices need not create equal impressions.
Keep a product-level record of regular price, actual selling price, promotion frequency and depth. Compare what a repeat buyer could have observed with what the new banner claims.
When comparing frequent shallow offers with rarer deep offers, use a longer randomized test and measure full-price buying after the promotion. The study itself measured remembered prices, not such a store test.
🎓 Findings
Alba and colleagues (1999) showed 61 students simulated shampoo price histories. In Study 5, both brands had the same actual mean price, $2.39, across 36 periods. For participants asked first to estimate the basket total, varying sale-price histories made the frequent-discount brand seem cheaper: $83.74 versus $86.68, a $2.94 perception gap (n = 14; one-tailed p < .02).
With only a regular price and a fixed sale price, the corresponding estimates were $84.74 versus $83.50; that contrast was not clear. The overall basket interaction was only marginal (p < .06). The average-price measure’s interaction was also not clear when all its respondents were included.
🧠 Why it works
People may remember a pattern of savings rather than calculate its exact average. Which pattern stands out depends on how variable the price history is.
✋ Limitations
This was a small classroom memory task with simulated histories. Its key interaction was marginal, and some average-price analyses changed after excluding modal-price answers. It cannot identify a profitable promotion calendar or a guaranteed frequency effect.
4. Follow customers after the discount ends
📈 Recommendation
Set discount depth using a full customer horizon. A larger initial order response can coexist with fewer or cheaper later purchases.
Track a stable assigned customer cohort through the sale and its aftermath, including people who did not redeem. Report cumulative contribution per assigned customer.
Separate existing customers from first-time customers. Choose a follow-up window that spans the category’s normal repurchase cycle, not just the campaign week.
🎓 Findings
Anderson and Simester (2004) randomly mailed deeper or shallower discount catalogues to established customers of a durable-goods seller. In Study A, 597 of 18,708 deeper-discount recipients bought, versus 761 of 37,758 controls: 3.19% versus 2.02%, or 1.17 percentage points more (p < .01). The two versions’ average sale prices on the 36 changed products were $77.17 versus $133.81.
Among those initial buyers, subsequent 28-month units averaged 6.89 versus 7.67, 0.78 fewer under the deeper offer (p < .01). Initial plus later units per buyer were not clearly different. Those future-buyer comparisons condition on who responded, so they combine selection into the sale with later behavior; they are not an effect on every mailed customer.
🧠 Why it works
A discount can move a planned purchase forward, change which shoppers respond or teach shoppers to wait for another deal. The paper examines these competing mechanisms rather than assigning the whole later difference to one cause.
✋ Limitations
The products were historical catalogue durables with an average repurchase interval near 48 weeks. Prospective-customer studies had different results, including a marginal repeat-unit result in Study B. This is not evidence that deep discounts always damage retention or that a particular percentage is optimal.
💡 The takeaway
A smaller price does not guarantee a better promotion.
Keep the offer honest, compare it with no promotion and measure cumulative contribution across the customers assigned to each test.
📚 Sources and study types
Cai, F., Bagchi, R., & Gauri, D. K. (2016; online 2015). Boomerang Effects of Low Price Discounts: How Low Price Discounts Affect Purchase Propensity. Journal of Consumer Research, 42(5), 804–816. Final author-upload full text; paid purchase and hypothetical-intention studies distinguished.
Alba, J. W., Mela, C. F., Shimp, T. A., & Urbany, J. E. (1999). The Effect of Discount Frequency and Depth on Consumer Price Judgments. Journal of Consumer Research, 26(2), 99–114. Final author PDF; classroom price-memory experiments with important marginal/null results.
Anderson, E. T., & Simester, D. I. (2004). Long-Run Effects of Promotion Depth on New Versus Established Customers: Three Field Studies. Marketing Science, 23(1), 4–20. Final author PDF; randomized catalogue depth trials and follow-up of initial responders.
