Seasonality and Black Friday: how not to overpay for the peak
In a peak both revenue and click prices rise, and not always at the same rate. How to prepare an account, when to raise budget, what to do a week before, and why the most important decision happens two months earlier.
- In a peak, demand rises and so does the cost of a click. What decides is whether margin on an order rises too — usually it does not.
- The most important work is done weeks ahead: measurement, feed, stock and data for automated bidding.
- Large step changes to budget mid-peak disrupt bidding strategies at the worst possible moment.
A peak season invites a simple line of reasoning: people buy more, so let us add budget. Except you are not the only one adding. Demand rises, competition rises and the cost of a click rises — while margin on an order typically falls during a discount period. Whether the peak earns for you is not decided in November. It is decided in September.
What actually happens in a peak
- Higher conversion rate. People are more decided and buy faster. This one works for you.
- Higher cost per click. Everyone is bidding up. This one works against you.
- Lower margin. Discounts, free shipping, gift wrapping. Most often forgotten.
- A shorter decision cycle. The conversion window compresses and attribution behaves differently from the rest of the year.
- More returns afterwards. For fashion and gifts a substantial item that only shows up in January.
The right question is not “how much do we add” but “what cost per order pays off at our seasonal margin”. That is usually a lower number than the rest of the year, not a higher one.
Preparation calendar
8 to 10 weeks ahead
- Calculate the seasonal margin. After discounts, free shipping and expected returns. Derive the target from it.
- Check measurement. Compare conversions with orders. Broken tracking costs most in a peak.
- Go through the feed. Disapproved products, availability, images. See feeds.
- Verify stock and logistics. Advertising out-of-stock goods during a peak damages the brand for a year.
4 to 6 weeks ahead
- Start raising gradually. Automated strategies need time. A jump on 27 November is the worst possible scenario.
- Build audiences. September and October visitors are your most valuable group in November.
- Prepare creative and copy. Including versions for the case where the promotion changes.
- Set a custom label for seasonal range. It lets you switch parts of the range on and off without rebuilding campaigns.
1 to 2 weeks ahead
- Freeze structural changes. New campaigns and strategy switches belong before this window, not inside it.
- Set seasonality adjustments where the platform offers them — they exist precisely so the strategy expects a different conversion rate.
- Check site speed and server capacity. A peak is the one time your site will fall over.
- Prepare a shutdown plan. What to switch off when stock runs out or logistics jams. That decision is harder to make on a Friday evening.
During the peak
- No large step changes. Add in smaller increments, even if it feels slower than you would like.
- Watch availability more often. Ideally continuously, not once a day.
- Do not panic about a higher cost of sale. It is expected. What matters is whether it holds against the seasonal margin.
- Watch disapproved products. Price changes during promotions are the most common cause of disapproval in exactly this period.
- Do not check the account every hour. It tempts interventions that, in aggregate, do more harm than good.
After the season
- Return budgets and targets gradually, not at once. The same logic as when raising.
- Wait for returns before evaluating the season. In fashion that can take a month and change the answer substantially.
- Evaluate against last year’s season, not against a November without discounts.
- Write down what did not work. Nobody will remember in a year, and next season gets planned from this one.
- Use the audiences you gathered. January customers from December purchases tend to be the cheapest source of repeat revenue.
Seasonality is not only Black Friday
Most fields have their own peaks that never appear in general articles: garden in spring, construction in summer, education around enrolment dates, travel in January and February, car servicing at tyre-change time. The same logic applies to all of them, and competition is often less prepared — so they pay off better than joining the November battle.
A practical step: take two years of data, split revenue by week and find your own peaks. Budget belongs to them, not spread flat.
What to take away
You do not win a season with budget, you win it with preparation. Work out your seasonal margin and derive the target from it — it usually comes out stricter, not looser, than the rest of the year. Raise in steps weeks ahead, freeze structural changes, and above all have a plan for what to switch off when something goes wrong. Evaluate only once the returns are in.
Common questions
When should I start raising budget before Black Friday?
Four to six weeks ahead and in smaller steps, not in a jump in the final week. Automated bidding strategies need time to find a new equilibrium after any larger change, and a big jump just before the peak is the worst possible timing. Structural changes such as new campaigns or switching strategy belong even earlier — one to two weeks before the peak the account should already be frozen.
Should I loosen the return target during a season?
Usually not, rather the opposite. Conversion rate does rise in a peak, but so does the cost per click and, above all, margin falls because of discounts, free shipping and higher returns. The target should therefore be derived from the seasonal margin, which tends to be lower than the usual one, and the result is often a stricter target rather than a looser one. A higher cost of sale during a peak is expected; what matters is whether it holds against the seasonal margin, not the annual one.
When should I evaluate the season’s results?
Only after returns, which for fashion and gift items can mean a month after the peak ends. An evaluation done immediately afterwards systematically overstates, because it counts orders that will come back. Compare against the same season last year rather than the preceding month, and make a point of writing down what did not work — nobody will remember in a year, and next season gets planned from this one.