CRO / Pricing
How to Run a Sale Without Training Customers to Wait
// a sale should be an event, not a schedule
By Fracto Solutions
October 12, 2026
The short answer
Make discounts occasional and tied to a real reason, keep them finite, and target new or lapsed buyers rather than everyone, because discounting affects existing and new customers in opposite directions. Measure incremental gross profit with a holdout, and keep urgency claims genuine: false ones are banned outright in the EU and UK.
Why do frequent sales backfire?
Because customers learn the pattern and your reference price moves. Kalwani and Yim found that both promotion frequency and discount depth lower the price customers expect to pay, and that the absence of an expected deal hurts more than an unexpected deal helps.
The damage is measurable over time. Mela, Gupta and Lehmann tracked 8.25 years of scanner panel data and found consumers became more price- and promotion-sensitive as promotion rose and advertising fell. Jedidi, Mela and Gupta put a number on the cost: a 5% increase in promotion frequency or depth reduced brand profit by 0.31% to 0.49%, negative across every brand they studied.
One important qualification, because the usual framing is too broad. Anderson and Simester’s three randomized catalog field experiments found deep discounts reduced established customers’ future purchases by about 10.8%, while increasing new customers’ future purchases by 14% to 34%. The habit forms among the people who already buy from you. That is precisely why targeting matters. This is the execution guide under the discounting pillar.
How do I keep sales from becoming a habit?
Make them rare, reasoned and finite, and tie each to a genuine occasion. On timing there is a real tension in the research worth understanding rather than a simple “be unpredictable” rule.
Krishna (1994) found customers buy larger quantities on deal when deal timing is more certain, and also when deals are spaced further apart holding frequency constant. That supports making your calendar hard to read. But Swait and Erdem (2002) found inconsistent promotion patterns lower brand utility and increase choice uncertainty, both of which reduce share. The two point in different managerial directions.
The reconciliation we use: be unpredictable in timing, consistent in presentation. A customer should not be able to forecast when the next sale lands, and should always recognize the brand running it. Chaotic promotion is not the goal.
Do
Avoid
Tie sales to a real occasion
Predictable monthly or weekly sales
Keep them time-bound and genuinely finite
An always-on "sale" section
Vary timing, keep brand presentation steady
A cadence customers can schedule around
Target new and lapsed buyers
Site-wide cuts your regulars see
Row two is a legal point as well as a brand one. See the rules section below.
How do I target sales instead of going blanket?
Aim discounts at specific jobs and segments: a welcome offer for new buyers, a win-back for lapsed customers, or a clearance on specific stock, rather than a site-wide cut everyone sees.
A blanket sale discounts customers who would have paid full price and shows everyone that waiting works. Targeted offers avoid both, and the Anderson and Simester split gives you the logic: discounting is a reasonable acquisition tool and a costly retention one. A welcome offer converts a first-time buyer, a win-back reactivates a lapsed one, and a clearance moves specific stock.
On clearance specifically, one correction to the usual advice: the break-even table in the pillar does not apply. Once stock is genuinely dead the original cost is sunk, so the right test is net realisable value against the alternatives, carrying cost, liquidation or write-off, and selling below cost is frequently the correct decision even though break-even arithmetic calls it impossible.
What are the rules on urgency and "limited time"?
This is the part most guides treat as a brand question and it is not. False urgency, false scarcity and continuously running “limited time” sales are blacklisted practices in the EU and UK, banned outright with no need to prove consumer harm. General information, not legal advice.
UK: DMCC Act 2024 Schedule 20 paragraph 7 bans “falsely stating that a product will only be available for a limited time, or that it will only be available on particular terms for a limited time, in order to elicit an immediate decision”, with paragraph 5 on false availability and paragraph 23 on “free” claims. CMA enforcement powers commenced 6 April 2025, with direct fines up to the greater of GBP 300,000 or 10% of worldwide turnover. EU: the same practices sit in Annex I of the Unfair Commercial Practices Directive. US: the FTC’s 2022 dark-patterns report names the “baseless countdown timer”, “false limited time message”, “false low stock message” and “false high demand message” by name, and private class actions are the larger live exposure.
The CMA’s lead case makes it concrete. Promotions presented as time-limited ran continuously for 10 and 15 months with no gaps; countdown timers counted to zero without any change in the underlying offer because a similar promotion reappeared immediately; and high-demand messages fired on as few as two website visits, with one product showing thousands of views against a handful of sales and ample stock. If you use urgency, keep contemporaneous evidence that the constraint was real.
On the sale itself: in the EU, Price Indication Directive Article 6a has required since 28 May 2022 that any announcement of a price reduction state the lowest price you applied in at least the previous 30 days, and the CJEU confirmed in 2024 that a percentage reduction must be calculated on that figure. In the US, FTC 16 CFR §233.1 requires a bona fide former price offered on a regular basis for a reasonably substantial period, and California adds a three-month prevailing-price test, under which HP settled a class action for $4 million in April 2025 over models marked down continuously.
How do I protect the full-price business?
Keep the full-price offer strong on its own merits so customers buy without waiting, and measure incremental gross profit rather than the sale spike. If you can only move product on discount, that is a demand or positioning problem a sale will not fix.
The honest scorecard is incremental profit, and the only way to get it is a holdout: keep a random slice of the audience out of the promotion and compare. Raw sales during a sale are misleading because some of those buyers would have paid full price and every unit gave away margin.
You will see “most of a promotion’s lift is just switching” quoted at around 75%. That was corrected: Van Heerde, Gupta and Wittink showed in 2003 that in unit sales terms only about 33% comes at the expense of other brands. Both figures measure cross-brand switching in CPG categories anyway, not the question you have, which is what share of your own discounted orders would have happened at full price. No published figure exists for that, and none could, because it depends on your store. Track full-price conversion and margin over time, the same discipline behind our growth-strategy pillar, and watch whether full-price sales dip before and after a promotion as customers time their purchases.
We design promotions that drive revenue without eroding full-price demand.
We keep sales occasion-based, finite, and targeted, protect the full-price offer, and measure full-price conversion and margin alongside the sale spike, so promotions add to the business instead of training your best customers to wait.
It’s the same measurement-first approach behind our conversion optimization and analytics work, where we rebuild the buying path and measure the result .
// we would rather show up in your reporting than in your inbox
Frequently asked questions
Why shouldn't I run regular sales?
Because frequency and depth both lower the price your customers expect to pay, and the absence of an expected deal then hurts more than an unexpected one helps. Over 8 years of panel data, customers became steadily more price-sensitive as promotion rose. Note this bites hardest on your existing customers.
Does discounting train all customers to wait?
No, and this is the most misquoted part of the topic. Anderson and Simester’s three randomized field experiments found deep discounts reduced established customers’ future purchases by about 10.8% but increased new customers’ by 14% to 34%. Discounting is a reasonable acquisition tool and a costly retention one.
Should sales be site-wide?
Usually not. A blanket sale discounts customers who would have paid full price and shows your regulars that waiting works. Target offers at specific jobs instead: new buyers, lapsed customers, or specific stock you need to clear.
Is an always-on sale section actually illegal?
It can be. False limited-time claims are banned outright under Schedule 20 paragraph 7 of the UK’s DMCC Act and Annex I of the EU’s UCPD, with no need to prove harm. In the CMA’s lead case, promotions presented as time-limited ran continuously for 10 and 15 months and countdown timers reset without the offer changing.
How do I know if a promotion actually worked?
Hold out a random slice of the audience and compare incremental gross profit, not the revenue spike. Some discounted buyers would have paid full price, and no published benchmark can tell you what share, because it depends entirely on your store and your customers.
Sources
Reference prices and promotion expectations: Winer, Journal of Consumer Research 13(2), 1986, established that choice depends on the gap between observed price and an internally held expected price. Kalwani and Yim, Journal of Marketing Research 29(1), 1992, found that both promotion frequency and discount depth lower the expected price, and that the absence of an expected deal hurts more than an unexpected deal helps. Blattberg, Briesch and Fox, Marketing Science 14(3 Supplement), 1995, generalize that deal frequency changes the reference price and a lower reference price reduces the premium a brand can charge. https://journals.sagepub.com/doi/10.1177/002224379202900108
Long-run cost of promotion: Mela, Gupta and Lehmann, Journal of Marketing Research 34(2), 1997, 8.25 years of scanner panel data, found consumers became more price- and promotion-sensitive as promotion rose. Jedidi, Mela and Gupta, Marketing Science 18(1), 1999, 691 households and 13,664 purchases: a 5% increase in promotion frequency or depth reduced brand profit by 0.31% to 0.49%, uniformly negative across four brands. Scope caveat: CPG scanner-panel data from single US markets in the 1980s and 1990s. https://people.duke.edu/~mela/Jedidi_Mela_Gupta_1999.pdf
The effect reverses by customer type: Anderson and Simester, “Long-Run Effects of Promotion Depth on New Versus Established Customers: Three Field Studies”, Marketing Science 23(1), 2004. Three randomized catalog field experiments covering 56,466 established customers and 542,026 prospects, discounts 40% to 47% deeper, followed 22 to 28 months. Deep discounts reduced established customers’ future purchases by about 10.8% while increasing new customers’ future purchases by 14% to 34%. The most rigorous direct-marketing evidence available, and it contradicts the universal version of “discounting trains customers to wait”. https://www.kellogg.northwestern.edu/faculty/anderson_e/htm/personalpage_files/Papers/Long_Run_Effects_of_Promotion_Depth_on_New_versus_Established_Customers.pdf
Predictability versus depth: Krishna, Marketing Science 13(4), 1994, found quantity purchased on deal is larger when deal timing is more certain, and larger when deals are spaced further apart holding frequency constant. Swait and Erdem, Journal of Marketing Research 39(3), 2002, found inconsistent promotion patterns lower brand utility and increase choice uncertainty. The honest synthesis: predictable timing lets customers optimize against you, while erratic promotion damages brand evaluation, so “be unpredictable” is not a safe blanket inference. https://pubsonline.informs.org/doi/10.1287/mksc.13.4.351
How much of a promotion is incremental: Gupta, Journal of Marketing Research 25(4), 1988, decomposed promotional elasticity as roughly 84% brand switching. That figure is widely misquoted as applying to sales. Van Heerde, Gupta and Wittink, “Is 75% of the Sales Promotion Bump Due to Brand Switching? No, Only 33% Is”, Journal of Marketing Research 40(4), 2003, showed that in unit-sales terms only about 33% comes at the expense of other brands. Both measure cross-brand switching in CPG categories, not what share of your own discounted orders would have happened at full price, for which no published figure exists. https://doi.org/10.1509/jmkr.40.4.481.19387
Urgency and scarcity claims: false urgency is blacklisted, banned per se with no need to prove harm. UK: DMCC Act 2024 Schedule 20 paragraph 7, “falsely stating that a product will only be available for a limited time… in order to elicit an immediate decision”, plus paragraph 5 on false availability and paragraph 23 on “free” claims; CMA enforcement powers commenced 6 April 2025 with direct fines up to the greater of GBP 300,000 or 10% of worldwide turnover. EU: the same practices are blacklisted in Annex I of the Unfair Commercial Practices Directive. US: the FTC’s September 2022 dark-patterns report names the “baseless countdown timer”, “false limited time message”, “false low stock message” and “false high demand message”. In the CMA’s lead case, time-limited promotions ran continuously for 10 and 15 months, countdown timers reset without any change in the underlying offer, and high-demand messages triggered on as few as two website visits. https://www.legislation.gov.uk/ukpga/2024/13/schedule/20 and https://www.ftc.gov/system/files/ftc_gov/pdf/P214800+Dark+Patterns+Report+9.14.2022+-+FINAL.pdf
Advertising the sale itself: EU Price Indication Directive 98/6/EC Article 6a, applicable from 28 May 2022, requires any announcement of a price reduction to state the lowest price applied in at least the previous 30 days; the CJEU confirmed in Case C-330/23 (Aldi Sued, 26 September 2024) that a percentage reduction must be calculated on that figure. US: FTC Guides Against Deceptive Pricing, 16 CFR Part 233; a former price must be bona fide and offered on a regular basis for a reasonably substantial period (§233.1), and §233.4 makes a “free” or two-for-one offer deceptive if the seller raises the price of the required purchase or cuts its quantity or quality. California Business and Professions Code §17501 adds a three-month prevailing-price test, and perpetual-promotion class actions under it are constant; HP settled one for $4 million in April 2025. https://eur-lex.europa.eu/eli/dir/1998/6/2022-05-28/eng and https://www.ecfr.gov/current/title-16/chapter-I/subchapter-B/part-233
Stated as Fracto’s position, not research: that sales should be occasion-based, finite and targeted rather than site-wide, that an always-on sale section is a mistake, and that clearance should be judged on net realisable value rather than on the break-even table. The first two follow from the reference-price literature without being directly measured. The clearance point is accounting logic: once stock is genuinely dead the original cost is sunk, so selling below cost is frequently correct even where the break-even table calls it impossible.