First, check the plan's number before blaming the sale
Start with the target, because a target that could never be reached makes every sale look like a failure. Find where it came from. A common source is last year's event copied forward, but last year may have run on different weekdays or over a holiday, with a bigger list, a wider range or more budget. The comparing periods guide aligns weekdays and moving holidays to the event's own dates and separates growth from seasonality; the measuring promotions guide builds a BaselineAn estimate of what a metric would have done without a given change, campaign or promotion, against which the effect is measured. several ways and reads the result under each. Then check what the plan counted. A plan in orders, a plan in revenue and a plan in contribution can disagree about the same sale; if orders met the plan but revenue did not, the gap sits in basket size, which the diagnosis for a falling average order value covers.
Next, make sure the window has filled. Ad platforms date a purchase by the ad interaction that led to it, so the last days of a sale keep filling in after it ends; the Meta guide to sales and seasonal peaks and the Google Ads guide to sales and tracking outages both read a sale only once delayed conversions have arrived. Use the order system as the record. Then split the shortfall as the diagnosis for a missed marketing forecast splits any miss, into volume, rate and timing.
| Where the gap shows | What it suggests | Causes to check first |
|---|---|---|
| Fewer sessions than planned, from every channel | Fewer people heard or cared | The list, the paid campaigns, demand spent before |
| Sessions as planned, fewer orders | Visitors met something that stopped them | The offer only partly live, the lead products out of stock |
| Orders as planned, less revenue | Smaller baskets or deeper discounts | The mechanic and eligibility, then order value |
| A strong first day, then a fade | The sale ran out of something | Stock of the lead products, the list's waves |
| Weak weeks before the sale as well as during it | Demand moved earlier or waited | Demand spent before, customers waiting for sales |
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| The sale beats a like-for-like baseline but misses the target | Where the target came from; last year's dates, list, range and budget | The target came from an unadjusted comparison | Rebuild the target from a like-for-like baseline and judge the sale against it |
| Shopping ads, emails or badges showed the wrong price; some items disapproved | Prices and codes in site, feed, ads and email on the first day; item status in Merchant Center | The offer never reached shoppers intact | One price everywhere, checked before launch and on the first morning |
| Paid costs rose and delivery was unstable through the window | Ad set creation and edit dates; learning status; budgets and limits | Campaigns launched or rebuilt inside the window, or budgets left at everyday levels | Launch early, freeze the structure, raise budgets in planned steps |
| Email orders were slow, unsubscribes rose and buyers got reminders | Send times by segment; buyer suppression; who heard first | The list heard late or all at once, and buyers were chased | Stage by engagement tier, early access for the best customers, remove buyers |
| Traffic arrived but few extra orders, mostly from regular customers | The job, mechanic, depth and eligibility; full-price share before past sales | The offer did not fit the promotion's job, or customers expected more | Match the mechanic to the job and protect full price between promotions |
| Orders faded midweek; sessions to the lead products held | Stock and variant availability of the lead products by day | The lead products sold out or were held back early | Set stock aside for the sale and shift, queue or stop demand when a line runs low |
| Weeks before the sale fell below baseline; codes redeemed outside their channel | Earlier promotions, pre-sales and live codes; units per order of existing customers | Demand was spent before the window | Space promotions, expire codes with their promotion, read pull-forward against a baseline |
Fix, verify and prevent each cause
A target built on the wrong comparison
A target copied from last year inherits everything that made last year different. The comparing periods guide compares like with like on weekdays, holidays and events, and the measuring promotions guide adjusts last year for trend and warns that last year may have carried its own promotion.
Fix: rebuild the target from a like-for-like baseline, adjusted for days, list size, range and budget, and judge the sale against that as well as the original number.
Verify: the rebuilt target and at least one other baseline, from the weeks before the announcement or from products the sale did not cover, tell the same story.
Prevent: write the baseline and its method into the plan before the sale, so the target and the judgement use the same comparison.
An offer only partly live
A sale price shown on the page but missing from the shopping ads, or a code in the email that fails at checkout, turns paid traffic away. The guide to merchandising for campaigns keeps the price identical on every channel, from the site and the feed to the ads and the emails, and records that products whose feed data disagrees with the landing page can be disapproved, as Google's product data specification sets out; sale prices must be in the feed from the moment they are on the page.
Fix: compare the price, code and badge for the lead products in every channel, correct the source, and check Merchant Center for items disapproved since the change.
Verify: the same price appears everywhere and the items show approved.
Prevent: run the sale promotion readiness checklist, which checks that the sale price is the same on the site, in the feed and in the ads and walks the checkout with a discount codeA code a customer enters at checkout to receive an offer, also used to identify sales from a creator, partner or channel that links cannot capture., before launch, then repeat the price check on the first morning.
Paid campaigns still learning when the sale opened
Meta's documentation on the learning phase, as the Meta peak guide cites it, says that a new or heavily edited ad set delivers unevenly, and costs more per result, until it has gathered enough optimisation events. An ad set built the night before a sale spends the window in the Learning PhaseThe period after launch or a significant edit when an ad set's delivery is still being optimised., and the guide notes that a large budget change can restart it. On Google, the outages guide prefers seasonality adjustments to targets changed by hand, and holds back strategy, conversion and structure changes during the event.
Fix: inside a running sale, stop editing: leave learning ad sets alone and keep further changes to the budgets the plan wrote down.
Verify: delivery steadies and cost per result settles as the ad sets exit learning.
Prevent: launch the peak structure early enough to settle, add sale creative to settled ad sets, and raise budgets in steps in the run-up, as the Meta guide advises.
A list told late or all at once
The tutorial on sale and launch emails decides who hears first, offers early access to the customers and subscribers most likely to act, stages the announcement by engagement tier, removes buyers before every follow-up and sends reminders only to people who showed interest. A single blast on the first morning skips all of that, and reminders that reach buyers annoy the people the next sale needs.
Fix: for the rest of the window, suppress buyers before every send and narrow reminders to people who clicked or visited without buying.
Verify: orders per email sent hold across waves and unsubscribes fall back.
Prevent: write the send order, the tiers and the suppression into the sale brief, with each wave's time.
An offer that did not fit its job
The guide to planning promotions gives each promotion one job, a mechanic and eligibility that fit it, and a frequency that keeps the full price believable. An offer to everyone is a poor way to win new customers, as the guide notes, and customers who have learned that the next sale is deeper wait for it, the opposite pattern the diagnosis for customers waiting for sales treats. Claims must be true: the planning guide notes that UK consumer protection law reaches reference prices never genuinely charged and untrue urgency, with the Competition and Markets Authority as the body that enforces it, and the guide to honest persuasion applies that to countdowns and stock messages; this is not legal advice.
Fix: do not deepen the discount mid-sale; record which job the offer failed and choose the mechanic for the next one.
Verify: the next promotion's incrementalConversions caused by marketing. orders against the baseline, not its headline orders, meet its break-even.
Prevent: work out the break-even uplift with the break-even calculator and the guide to contribution margin and break-even before approving any mechanic.
Lead products that ran out
The products a sale features are the ones its ads and banners send people to. The guide to marketing within capacity counts demand the business cannot serve as wasted spend, a damaged experience and distorted data, and offers four responses: shift demand to lines with room, slow it, queue it, or stop it precisely. A stock-outA period when a product is out of stock and cannot be sold, which caps sales and should stop advertising for that product until stock returns. in a popular size or colour midweek sends the sale's visitors to an unavailable item.
Fix: move the next available lines into the lead places, pause the ads that feature the missing variants and offer a waitlist where stock is coming.
Verify: sessions to unavailable variants fall and the sale's conversion recovers.
Prevent: agree before launch how much of each lead product is set aside for the sale and what to do as a line runs short, as the merchandising guide advises. If a lead product was already selling less in the weeks before the sale, the diagnosis for a best-selling product's sales falling finds what it had lost.
Demand spent before the window
The measuring promotions guide reads 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., stock-up and cannibalisation against the baseline. An earlier promotion, a pre-sale or a code still live on a voucher site can serve the sale's buyers in the weeks before it, and customers who stocked up do not need to buy again yet.
Fix: find the codes and offers live before the sale and expire any past their end time through the SOP for discount code requests, which keeps a register of who each code was issued to.
Verify: redemptions of old codes stop and the weeks before the next sale hold at their baseline.
Prevent: space promotions on the calendar, and close every offer in every channel at its stated end with the SOP for ending a sale or promotion.
Prevention: a written baseline, a readiness date and a mid-sale read
Most underperforming sales fail in ways the business could have seen coming. Three habits catch them early.
A written baseline. Before launch, write down how the sale will be judged: the baseline, its method, what is counted and when the readout happens. A target set from that baseline cannot be the wrong comparison, and a result read against it cannot be argued about afterwards.
A readiness date. The backward planning framework fixes the sale date and works back to the latest start of each step, so prices, the feed, the ad sets, the email waves and the stock set aside are ready by a date that leaves room to fix them. The peak season trading playbook runs the whole programme around that date, and the readiness checklist gates it.
A mid-sale read. Check each layer on the first morning and daily after, and change only what the plan allows.
| Layer | Check during the sale | Owner |
|---|---|---|
| Offer | The same price, code and badge on site, feed, ads and email; item approvals | Trading lead |
| Paid | Learning status and delivery; spend against the planned budget | Paid media owner |
| Waves sent on time; buyers suppressed; reminders only to interested people | Email owner | |
| Stock | Lead products and their variants available; replacements ready | Merchandiser |
| Codes | Redemptions by channel; codes appearing outside their channel | E-commerce manager |
Do not judge the result from these reads; they find breakages. Judge the sale once the window has filled, against the written baseline, with the measuring promotions guide, and record the layer that failed in the campaign post-mortem template so the next plan starts from it. Other store symptoms are collected in the e-commerce troubleshooting section.
A luggage shop's spring sale falls short (illustrative)
Hevrastoq Luggage finds where its spring sale went
Hevrastoq Luggage, an invented online luggage shop, plans 1,850 orders from a five-day spring sale and takes 1,120. The target was last year's sale copied forward, but last year's ran over a four-day bank holiday weekend and went to a list of 42,000 against this year's 31,000. Rebuilt for the same weekdays and the smaller list, the like-for-like baseline expects about 1,450 orders, so a shortfall of about 330 is real.
Checking the first day finds the feed still carrying full prices for 26 hours, so shopping ads showed no reduction and Shopping revenue on day one was a third of day two. The main cabin suitcase was out of stock in 2 of its 5 colours by Wednesday, while those colours still drew 38% of its sessions. The social ad sets had been rebuilt the night before and spent the first 3 days in learning, at £41 per purchase against £24 planned. And a 15% code from a winter partner offer was still live on a voucher site, with 210 redemptions in the four weeks before the sale.
For its autumn sale the shop sets the target from a like-for-like baseline, fixes a readiness date 10 days out for prices, feed and ad sets, sets aside 400 cabin cases in every colour, and expires partner codes with their promotion. The autumn sale reaches 97% of its rebuilt target.
Common mistakes
- Copying last year's number. Different weekdays, holidays, list size or budget make it the wrong comparison.
- Judging the sale the morning after. Delayed conversions keep arriving; read once the window has filled.
- Rebuilding ad sets the night before. They spend the sale in learning.
- Mailing everyone at once. Stage by tier, give the best customers early access and suppress buyers.
- Deepening the discount mid-sale. It raises the break-even and teaches customers to wait.
- Extending the deadline. It makes the original claim untrue.
- Leaving partner codes live. They spend the sale's demand before it starts.
- Reading headline orders as success. Only orders above the baseline, after pull-forward, are extra.
