A promotion needs a job
The first question about any promotion is not how deep the discount should be but what it is for. A promotion without a job tends to become a habit: the month looks slow, a code goes out, orders rise, and nobody asks whether those orders were extra or simply cheaper.
Most promotions do one of five jobs, and each job implies a different audience, mechanic and measure.
| Job | What success looks like | Usual audience | Usual mechanics | Judged on |
|---|---|---|---|---|
| Acquire | New customers who go on to buy again | People who have never bought | First-order offer, starter bundle, gift with first purchase | New customers won, and their second orders |
| Reactivate | Lapsed customers buying again | Customers who have not bought for longer than usual | Personal offer, members' price, a reason to return | Lapsed customers who return, against a group left alone |
| Clear stock | Stock that will not sell at full price turned into cash | Bargain-seekers and existing customers | Clearance pricing on named lines | Cash recovered against the alternative of writing off |
| Lift the basket | Larger orders | Everyone, at the point of purchase | Spend threshold, bundle, multi-buy, free delivery above a level | Contribution per order, not order value alone |
| Mark a seasonal moment | Sales in a period when buyers expect offers | Everyone | Event sale, gift offers, early access | Contribution over the event and the weeks around it |
A promotion that tries to do several jobs usually does none of them well. An offer to everyone is a poor acquisition offer, because most of the people using it already buy; a clearance sale that includes the core range is a discount on the range, not a clearance. If a promotion seems to need two jobs, it is usually two promotions.
Writing the job down also fixes how the promotion will be judged. The measuring guide's scorecard reads a promotion on incrementalConversions caused by marketing. orders, margin given away and who it brought in; a written job says which of those numbers matters most for this promotion.
Choosing the mechanic
The mechanic is how the offer works: what the customer gets, and what they have to do to get it. Different mechanics cost different amounts and change behaviour in different ways.
| Mechanic | Best suited to | What it costs | Main risk |
|---|---|---|---|
| Percentage off | Seasonal events, clearance | The discount on every order in scope, including orders that would have happened anyway | Trains customers to wait; signals the full price is negotiable |
| Money off | First orders, reactivation | A fixed amount per order, which bites hardest on small orders | Small baskets that barely clear the minimum |
| Spend threshold | Lifting the basket | Margin or delivery cost on the orders that reach the threshold | Set too high it does nothing; set too low it gives margin away |
| Bundle or multi-buy | Lifting the basket, introducing a product | A smaller discount on the extra item | Cannibalising full-price sales of the parts |
| Gift with purchase | Acquisition, seasonal events | The cost of the gift, usually less than its perceived value | A gift nobody wants |
| Free delivery | Removing a barrier to the first order or a bigger basket | Delivery cost per order | Becoming expected on every order |
| Members' price or early access | Reactivation and loyalty | The discount, limited to people who have joined | A membership that exists only to collect codes |
The average order value levers guide sets out the basket-building mechanics in detail and the risk that comes with each. It notes, for instance, that a minimum basket for a promotion trains customers to wait for codes, and that a deep bundleSeveral products sold together as one item at a single price, used to raise order value; a kit is a bundle designed as a starting point for new customers. discount is a sign the fit between the products is weak. For spend thresholds and free delivery, the free shipping threshold test judges a new threshold on contribution per visitor rather than on AOVAverage revenue per order., which is the right measure for any basket-lifting promotion.
For a range sold in versions, the good-better-best framework suggests promoting in a way that keeps the tiers honest: an offer that makes the top tier cheaper than the middle one, even briefly, teaches buyers that the structure is arbitrary.
Who should be eligible
The most expensive part of most promotions is the discount given to people who would have bought anyway. The measuring guide calls this cannibalisation: people who would have paid full price use the code because it was there. Eligibility is the main lever for limiting it.
- New customers only. Suits acquisition. A first-order code that works only once per customer, checked against the order history, limits use by existing customers.
- Lapsed customers only. Suits reactivation. Define lapsed from how often customers normally buy, not from a round number, so the offer reaches people who have actually drifted rather than those between ordinary orders.
- Members or subscribers. Suits loyalty and early access. The offer rewards a relationship rather than a visit.
- Everyone. Suits seasonal events and clearance, where the job is volume or cash and the audience cannot be separated in advance.
Even a targeted promotion includes some people who would have bought anyway. The question is how many, and the answer comes from a comparison: a group that was eligible but not sent the offer, or a baseline built from comparable periods, as the measuring guide sets out. The guide's warnings apply at the planning stage too. Pull-forwardPurchases a promotion moves earlier in time rather than creates, as customers who would have bought later at full price buy during the offer. is demand borrowed from the weeks after the sale, and the weeks before it can fall when a sale is expected. A promotion aimed at regular customers is the one most likely to borrow from their next order, so plan the measurement window to include the weeks around it.
For lapsed customers, the win-back offer test compares an offer, a message without an offer and a holdout, which separates the value of getting in touch from the value of the discount. If a reminder alone brings back most of the customers who would have returned, the discount was buying very little.
How often: protecting the full price
Every promotion teaches customers something about the price. Occasional promotions with clear reasons teach them that the full price is normal and the offer is an event. Frequent promotions teach them that the full price is a starting point.
The pricing and packaging guide describes how this happens with discounts in sales-led businesses: each discount is reasonable in isolation, together they become the real price, and the list price becomes a fiction that buyers learn to ignore. The same applies to retail promotions. The campaign calendar guide adds that promotions lift conversion while live, depress it afterwards, pull demand forward and train customers to wait.
The signs that promotions have become too frequent are visible in the business's own data:
- A rising share of orders placed with a discount, and a falling share at full price.
- Weak weeks between promotions, especially just before a promotion customers have learned to expect.
- Customers asking when the next sale is, in reviews, support messages or social comments.
- New customers whose first order was discounted buying again only when another discount appears, which shows in a weaker Repeat Purchase RateShare of customers who placed more than one order in the period. at full price.
The remedy is fewer promotions, each with a job, and stretches of full-price trading between them that are deliberately protected: no codes, no sitewide offers, and full-price messages that give a reason to buy now that is not a reduction. Where the team doubts that fewer offers will pay, the full-price protection test is a design for comparing the current offer rhythm with offers at planned events only, on randomly assigned customers.
Honest prices, honest deadlines
Promotions are where price claims are most often made, and the rules on them matter. The pricing and packaging guide records that in the UK the Competition and Markets Authority enforces consumer protection law covering misleading pricing practices, including reference prices that were never genuinely charged, urgency claims that are not true and fees revealed late in the purchase, and that the Advertising Standards Authority's advertising codes cover price statements in advertising, including claims about savings and comparisons with other prices. Other markets have their own rules, and the direction is similar.
In planning terms that means three checks before any promotion is briefed:
- A "was" price must have been a real price. The reference price should be one customers were actually charged, and the evidence for it should be on file before the promotion goes live.
- A deadline must be real. If the offer says it ends on a date, it ends on that date. A sale that is extended or repeated straight away makes the original deadline untrue.
- The advertised price must be the price people can pay. Exclusions, minimum spends and charges belong in the offer, stated clearly, not discovered at checkout.
The sale promotion readiness checklist checks that reference prices and discount claims can be substantiated before the day. None of this is legal advice; promotions with reference or comparative pricing should be reviewed by whoever handles compliance before they run. On the site itself, countdown timers, stock messages and pre-selected add-ons need the same checks, which the guide to honest persuasion sets out with the returns and complaints to watch when testing them. Keeping the deadline real at the other end is the job of the SOP for ending a sale or promotion, which takes the offer out of every channel at the end time.
Setting the break-even before the promotion
A discount lowers the Contribution MarginRevenue left after every cost that varies with each sale. on every order it applies to by the full amount of the discount, because the product, delivery and fees cost the same. So a promotion has to sell more just to leave the business where it started. Work out how much more before deciding to run it.
Break-even sales uplift = Discount ÷ (Contribution margin − Discount), with both expressed as shares of the full price(ratio)
The shape of the result matters more than any single figure. When the discount is small relative to the margin, the uplift needed is modest. As the discount approaches the margin, the uplift needed grows very fast, and a discount equal to the margin can never pay back on the orders it touches, however many there are.
Then add the promotion's own costs, the media, production and any extra fulfilment, and use the break-even calculator with the discounted order value and the margin left after the discount: it returns how many orders are needed to cover that spend. Compare the total with what the business can realistically expect from its own past promotions, read after pull-forward and cannibalisation, not from the headline. If the comparison is uncomfortable, change the mechanic, narrow the eligibility or drop the promotion.
When the right depth is genuinely uncertain, test it rather than guess. The discount depth test splits recipients into a deeper discount, a shallower one and no offer, and compares them on gross profit per recipient after returns. If a promotion planned this way still sells well below its plan, the diagnosis for a sale or promotion underperforming checks whether the target, the offer's delivery in each channel or the offer itself fell short.
Fitting promotions into the calendar
Promotions belong in the campaign calendar as one of its layers, planned for the year rather than added when a month looks slow.
- Place the seasonal moments first. Events customers expect, and the SeasonalityPredictable, repeating changes in demand tied to the calendar, such as months, holidays, weekdays or hours. of the business's own demand, set the fixed points.
- Give each promotion its job and its weeks. Space promotions so that full-price stretches sit between them, and avoid running two promotions to the same customers at once.
- Plan the weeks around each promotion. Expect a dip afterwards and possibly before, and do not judge the next campaign against them.
- Prepare the pages early. For recurring sale events, the seasonal sale pages guide recommends one permanent address per event that keeps its links from year to year, prepared well before the peak.
- Prepare the bidding. Automated bidding learns from recent conversion rates, so a short sale can mislead it. The guide to bidding through sales and tracking outages covers seasonality adjustments, budgets for the peak days and the changes to hold back while the sale runs.
- Run the readiness check. The readiness checklist covers the offer, site, stock, media and owned channels in the days before prices change.
- Brief affiliate partners early. Partners publish the offer on their own pages and schedules, and the guide to running promotions through affiliates covers their offer calendar, codes, rates and clean-up.
- Freeze risky changes around the peak. The tutorial on planning a change freeze stops tag, template and automation changes for the period while posting and scheduled sends carry on.
- Brief the support team. Customers ask about the offer as soon as they see it, and the tutorial on briefing customer service before a campaign gives agents the terms, the claims and what they may offer before the first question arrives.
The other documents in the planning section, the brief and the budget, should carry each promotion's job, eligibility and break-even, so the people building it know what it is for. The promotion brief template puts those decisions, with the prices, deadline, stock, channels and end plan, on one page that every owner builds from.
Alternatives to a discount
Many promotion jobs can be done without cutting the price, and the alternatives often protect the full price better.
| Job | Alternatives that cost no discount |
|---|---|
| Acquire | A starter kit priced fairly, a guarantee or easy returns, samples, content that answers buyers' questions, referrals |
| Reactivate | A reminder or a message about what is new, a replenishment prompt, a personal recommendation |
| Clear stock | Bundling the line with a best-seller, selling through a separate outlet channel, a gift with purchase |
| Lift the basket | Curated cross-sells, a well-placed threshold for delivery, kits that save the buyer effort |
| Mark a seasonal moment | Gift guides, limited editions, early access for members at full price, events |
The win-back test's message-only arm is a direct check on the first of the reactivation alternatives. More generally, if the business could do the job with a message, a product or a service improvement, the discount should have to earn its place against that option.
A yarn shop rethinks its sales calendar (illustrative)
A yarn shop replaces monthly sales with four planned promotions
Skeinbright Yarns, an invented online yarn shop, ran a sitewide 20% sale in 9 months of the last 12. Its contribution margin at full price is 48%, and 61% of last year's orders used a discount. Weeks just before a sale were its weakest, and customers in its reviews asked when the next one would be.
The owner replaces the monthly sales with four promotions, each with one job, and works out the break-even uplift for each before booking it. For a 20% discount on a 48% margin, the shop needs 20 ÷ (48 − 20), about 71% more sales, just to stand still.
| Promotion | Job | Mechanic and eligibility | Weeks | Break-even |
|---|---|---|---|---|
| Spring spend threshold | Lift the basket | Free delivery and a free pattern above £60, everyone | 2 weeks in March | Each order reaching £60 must add at least £9 of contribution |
| Members' early access | Reward loyalty | 15% off new season yarns for 3 days before launch, members only | 3 days in September | 45% more sales from members than a normal launch week |
| Discontinued colours | Clear stock | 40% off 64 colours being dropped, everyone | 2 weeks in January | Cash recovered must beat the £1,900 a wholesaler offered |
| Lapsed-customer offer | Reactivate | £8 off a £40 order, customers with no order in 9 months | Rolling, by email | Returning customers must exceed a no-offer holdout by enough to cover the £8 |
Between them, the offers open to everyone cover about 5 weeks of the year; the other 47 carry no sitewide codes, and the lapsed offer reaches only customers who have stopped buying. It runs with a message-only group and a holdout, as in the win-back test. Over the following year discounted orders fall from 61% to 27% of the total, full-price weekly sales rise by about a fifth, the clearance recovers £2,600 against the wholesaler's £1,900, and the message-only group returns at nearly the same rate as the offer group, so the £8 offer is cut to £4 for the next year.
Common mistakes
- Running a promotion without a job. Without a job there is no right mechanic, audience or measure.
- Offering everyone the discount. Most of the people who use an open code would have bought anyway.
- Discounting so often the full price stops meaning anything. Customers learn to wait, and the weeks between promotions weaken.
- Choosing the depth without a break-even. As the discount approaches the margin, the sales needed to stand still grow very fast.
- Quoting "was" prices that were never charged. A reference price should be real and on file; deadlines should be real too.
- Judging the promotion on the headline. Pull-forward, stock-up and cannibalisation are part of the result; read the weeks around it.
- Leaving the bidding and the pages until the last week. Sale pages and bidding plans need preparing before demand rises.
- Never trying an alternative. A reminder, a bundle or a better first-order kit can do the same job without a discount.
