Black Friday and the holidays: the decisions are made before, not during
During a sales peak, most of the profitable decisions are already made or already lost. Any significant change to a campaign sends it back into learning, and a system relearning during your three best days spends without knowing who it is buying. The calendar matters more than the discounts.
- Meta documents that an ad set exits the learning phase after roughly 50 results in the week following the last significant change. Every edit resets the counter to zero.
- A campaign launched three days before the peak will spend the entire peak in learning, at the most unstable cost per result of the year.
- The most profitable decision of the season is a date: the one after which nothing more is touched. It costs nothing to make.
- The checkout is the only place where a fix multiplies by the volume of the peak. Baymard puts the achievable gain from already documented frictions at 35.26%.
- A campaign can spend up to twice its average daily budget on a given day. That is a useful flexibility during a peak, provided you planned for it.
On this page
Freeze date
The freeze date is the day after which no further change is made to campaigns, budgets, landing pages or conversion definitions, until a period of high commercial activity ends. It exists because automated delivery systems relearn after every significant change, and a system relearning during a peak buys badly at the moment traffic costs the most. Setting it is a management decision, not a technical setting.
What a peak really amplifies
The usual conversation about these periods is about the discount: how much to offer, when to announce it, how to stage it. That is the visible part, rarely the one that decides the outcome.
A sales peak amplifies three things, and only three. It amplifies volume, so every friction in the buying path costs more than it normally would. It amplifies the cost of acquisition, because all your competitors are buying the same attention in the same week. And it amplifies your steering errors: a poorly set campaign burns in three days what it would burn in three weeks the rest of the year.
It does not amplify the quality of your offer, the clarity of your proposition or the soundness of your conversion tracking. Those three are locked in the moment the peak begins, and no last-minute setting fixes them. The season is therefore won in September and October, not during Black Friday week.
The right question to ask in a planning meeting is not "what discount do we offer." It is "what will we no longer be able to fix once the peak has started." The answer to that question gives you the list of projects to fund now, in order.
A ticket on-sale follows a different logic, that of a finite inventory sold in a few hours, covered in our article on the on-sale day peak.
What to settle before committing a peak budget
The peak concentrates a large share of many companies' annual revenue into a few days. That is what makes improvisation so costly: an error that would go unnoticed in March is paid for here in weeks of revenue.
- What share of our annual revenue rides on this period, and have we put a number on it?
- What margin do the products the discount will sell carry, once the discount is deducted?
- What is our freeze date, and who has the authority to lift it?
- Has our checkout been tested on mobile since the last peak?
- If performance collapses on the second day, what do we stop funding, and who decides?
The useful answer starts with the margin after discount and proposes a freeze date before talking about ad creative. A weak answer promises to optimize live during the peak, offers to launch the campaigns the week before, or talks about discounts without having seen your costs.
Deciding what has to be ready before the peak, and in what order to fund it, is settled on your real margin and your real calendar. A 90-minute consultation settles it, with a written summary your team or your agency can execute.
The freeze date, the cheapest decision of the year
This is the most profitable recommendation in this article and it costs nothing to apply. Set a date after which no one touches the campaigns, the budgets, the landing pages or the conversion definitions.
The reason is documented. Meta writes that changing certain details of a campaign, an ad set or an ad can restart the learning phase. It also states that you should wait for that phase to end before making changes, because performance is less stable during it and the results obtained then do not predict the ones that follow. Editing during the peak therefore means deliberately plunging the system back into its most unstable period, at the moment traffic costs the most.
Yet the opposite reflex is universal. On the second day, the cost per sale rises, someone gets worried, a bid is adjusted or a visual is swapped. The counter resets to zero, the instability worsens, another adjustment follows. The spiral costs more than the problem it claimed to fix.
| Timing | What is decided | What can no longer be decided |
|---|---|---|
| Two to three months before | Offer, margin after discount, inventory, budget thresholds | Nothing is locked yet |
| Six weeks before | Checkout fixes, landing pages, conversion tracking | Heavy site projects |
| Three weeks before | Campaign launch, final creatives | Account structure |
| Freeze date | Nothing more | Everything |
| During the peak | Stop a failing campaign, add budget to what works | Edit, test, replace |
| After | Reading results, decisions for next year | Making up for what was missed |
The "during the peak" row deserves a note. Stopping a campaign that produces nothing stays possible and sometimes necessary. Raising the budget of an ad set that works is too, but Meta notes that frequent budget changes can trigger a return to learning. The working rule: one substantial decision, made once, beats ten adjustments.
Why launching late dooms the campaign
This is the most frequent and most mechanical mistake. A new campaign launched a few days before the peak will cross the entire peak in the learning phase.
The math is simple. Meta documents that an ad set leaves the learning phase after roughly 50 results in the week following the last significant change. A business that produces thirty sales a week in normal times will not reach that threshold before the peak is over. The campaign will therefore spend its budget in its most expensive and least stable period, then stabilize in December, when demand falls back.
The practical consequence is a calendar, not a trick. Launching three weeks before the peak leaves two full weeks to clear the threshold, plus a week of margin. On a low-volume account, you have to go back further, or give up new campaigns and raise the budget of campaigns that are already stabilized.
Multiplying ad sets for the season makes exactly this problem worse. Meta writes that the more ads and ad sets you create, the less its system gathers information on each one, and recommends combining similar ad sets. Splitting the peak into eight segments guarantees that none will reach the threshold. The subject is developed in Facebook targeting.
On the search-engine side, the constraint is different but it exists too. Google states that a campaign can spend up to twice its average daily budget on a given day, without ever charging more than that budget would have allowed over a 30.4-day cycle. During a peak, this flexibility is good news: it absorbs an exceptional day. You just have to have anticipated it in your cash flow rather than discovering it on the statement.
The checkout, where the effect multiplies
This is the only project on this list whose effect multiplies mechanically by the volume of the peak, and it is the one most often put off.
The Baymard Institute measures an average cart abandonment rate of 70.22%, computed across fifty studies, and puts at 35.26% the conversion gain available to a large online store by fixing only usability problems it has documented as solvable. It also measures an average of 23.48 form elements displayed by default in a checkout, when a reduction of 20 to 60% is possible on most sites.
Apply these proportions to three days that concentrate a large share of your annual revenue, and the conclusion is clear. A point of conversion gained in October is worth, in November, several times what it would be worth in March. It is also the only gain that stays acquired after the peak, unlike the ad budget.
Fees discovered late in the checkout
It is the most cited cause of abandonment, and it is fixed by showing shipping fees earlier. The cost of the fix is one day of work. The cost of inaction is counted in lost carts multiplied by the volume of the peak.
The number of fields
Every field removed is one less obstacle between a decided buyer and their order. In a period when people buy fast and on mobile, the effect is more pronounced than the rest of the year.
Forced account creation
It blocks a buyer ready to pay, often a first-time buyer, that is to say exactly the customers a peak brings in. Guest checkout is set up once and pays off every year.
The site holding up under load
A site that slows down on the day of the peak turns an ad budget into pure expense. It is the only line on this list whose failure is total rather than gradual, and it is tested before, not during.
What stays in-house: the margin after discount per product, the real inventory and its depth, the company calendar, and the authority over the freeze date. That last one is a management decision: with no name attached, someone will end up touching the campaigns on the second evening. What gets delegated: account structure, campaign preparation and their launch on time, checkout fixes, load tests, tracking setup and reading the results. A business that freezes its calendar six weeks out gets a better return than a business that doubles its budget the day before.
What you can read during, and what has to wait
During the peak, almost nothing is readable, and that is the main source of regretted decisions.
| What you look at | Readable during? | Decision allowed |
|---|---|---|
| The site responds and the checkout works | Yes, continuously | Fix an outage, always |
| A campaign spends nothing at all | Yes | Check a rejection or a block |
| The cost per sale on a single day | No, too volatile | None |
| The comparison between two creatives | No, learning in progress | None |
| The overall return of the period | After the full period | Renew, adjust, stop |
| The real margin after returns | Four to eight weeks after | Next year's budget |
The last row is the one almost no one waits for, and yet it is the only one that says whether the season was good. Post-holiday returns change the result of a discount period, sometimes completely. A management team that stops its judgment on December 31 is judging an incomplete figure, and renews the following year a strategy it believes to be winning.
A useful note on attribution: do not change your attribution model during or just after the period. Comparing two months measured with two different models measures only the change of model, as detailed in reading results without a full model.
The peak does not reward those who react fast. It rewards those who have nothing left to decide.
Falia analysis gridChecklist six weeks before the peak
The break-even calculation is detailed in the move from reported ROAS to actual profit, the reliability of tracking in the three gaps that distort measurement, checkout fixes in the order of conversion fixes, and the time a test needs in Facebook ad testing.
The framework for choosing and concentrating budgets is set out in choosing and concentrating advertising budgets.
Turning a sales peak into a predictable result rather than a gamble is at the heart of the Optimize the profitability of your digital campaigns goal.
Already running a marketing team? See how we plug in as reinforcement on paid advertising.
Frequently asked questions about Black Friday and the holidays
When should you launch your peak campaigns?
At least three weeks before. Meta documents that an ad set leaves the learning phase after roughly 50 results in the week following the last significant change. A campaign launched a few days before will therefore cross the entire peak in learning, at the most unstable cost per result of the year.
Can you adjust your campaigns during the peak?
As little as possible. Meta states that changing certain details can restart the learning phase, and recommends waiting for that phase to end before making changes. Stopping a failing campaign stays possible. Editing, testing or replacing during the peak costs more than the problem you think you are fixing.
What is a freeze date?
The day after which no one touches the campaigns, the budgets, the landing pages or the conversion definitions. It is the cheapest and most profitable decision of the season, provided a named person has the authority to lift it, failing which someone will end up stepping in on the second evening.
Should you create several campaigns for the season?
Rarely. Meta writes that the more ads and ad sets you create, the less its system gathers information on each one, and recommends combining similar ad sets. Splitting the peak into several segments guarantees that none will reach the learning-exit threshold.
What should you invest in if the budget is limited?
In the checkout. It is the only project whose effect multiplies by the volume of the peak and stays acquired afterward. The Baymard Institute puts at 35.26% the conversion gain available by fixing only frictions it has documented as solvable.
When can you judge whether the season was good?
Four to eight weeks after, once the returns window has passed. Post-holiday returns change the result of a discount period, sometimes completely. Judging on December 31 amounts to renewing the following year a strategy evaluated on an incomplete figure.
- Meta, Business Help Center, "About the learning phase," on the exit threshold, the effect of ad-set volume and budget changes, accessed July 2026.
- Meta, Business Help Center, "How to edit Facebook and Instagram ad campaigns in Meta Ads Manager," on the changes that restart the learning phase, accessed July 2026.
- Google Ads Help, About target ROAS bidding, source of the rule that daily spend can reach up to twice the average daily budget, accessed July 2026.
- Baymard Institute, Cart abandonment rate statistics, average computed across 50 studies and checkout usability research, accessed July 2026.
Gabriel almost always takes your first call and carries out your audit. He builds the strategy starting from your growth goal: where to put your budget, which market to test and how to connect each lead to a real sale in your CRM. He mainly leads engagements for three goals: Optimize the profitability of your digital campaigns, Develop a new market, and Generate demand and growth. With Geneviève, he also works on organic search (SEO), AI visibility (GEO) and conversion rate optimization (CRO). The sales a Google Ads or Meta Ads campaign brings in depend on the page that receives the click. He writes mainly about marketing strategy, paid advertising and measurement.
About Falia →
SEO
Paid advertising
Conversion
AI visibility
Web design