Black Week with Google Ads: work out the discount first, then the campaign

The benchmark is Black Week without a discount

The biggest lever of Black Week, the week around Black Friday, is the discount, and you measure it against your own Black Week without a discount. An ordinary month is no comparison: the choice is between the same days with or without a discount.

A worked example with round, invented numbers: net price €100, variable costs €60, the same clicks and click prices in all three cases, €100 of ad spend each.

The same promotion is therefore either a mistake or a gain, depending on how much demand you keep without a discount. ROAS does not help here: the highest value in the table does not belong to the best result.

How many extra sales a discount has to bring

Every discount has a price in orders, and that price rises faster than the discount. With P as the net price, k as variable costs and r as the discount:

Required increase in orders = (P − k) ÷ (P × (1 − r) − k) − 1

Set this increase against what discounts actually do for you. The best way is to compare discounted and undiscounted products over the same days last year. If you lack that, run a comparable group without a discount this year. Both remain rough, because discounted products are rarely chosen at random. Which costs belong in k is covered in calculating break-even ROAS.

For a sense of scale: in a vendor study by Optmyzr, accounts without a seasonality adjustment saw conversion rates 7.5 to 17.5% above a comparison week during the promotion week in 2022 to 2024, and click prices 16 to 17% higher. The study does not split results by depth of discount. Measured against that, even a 10% discount in the example does not pay for itself through conversion rate alone. In my assessment, a discount therefore mainly pays off when it saves demand that would otherwise fall away, or when it brings additional clicks. Doubling the budget recovers the €100 in the 20% case, but only at the same efficiency, and that is the upper limit. The calculation also counts Christmas purchases that were simply brought forward as success.

Your stock position decides the discount

The answer comes from your stock and from how comparable your products are, not from the calendar. This is how I would decide:

Staying out does not mean switching off. The ads keep running with the normal target, and in my assessment rising click prices then slow spending through the target. The rule does not apply if you can include a proven new customer value: then the single order may look worse.

The target follows the margin after discount

The lower limit for target ROAS is P × (1 − r) ÷ (P × (1 − r) − k − g), where g is the amount that should remain per order after advertising. With a 20% discount and g = €5, the example gives 5.3. If your target is already above that, leave it. Below it you miss your minimum amount, and below 4.0 you are heading for a loss. Suppose your target is 2.5: €100 of ad spend brings €250 of revenue, just over 3 orders and a loss of around €38.

Google advises adjusting targets a few days before the promotion day, gradually, for example in steps of 30% with one to two conversion cycles in between. Going from 2.5 to 5.3 would take three such steps, and the higher target already throttles full-price sales during the run-up. If that is too expensive for you, the discount is too deep. At 10%, the lower limit is only 3.6.

For Search and Performance Max there is also Promotion Mode as a beta: you schedule a temporary change to the ROAS tolerance and additional daily budget in advance. Google does not mention Standard Shopping here. The tolerance sets how far bids may drop below the target. It does not remove the lower limit, so I would only open it as far as the gap to break-even after discount allows.

The formula assumes that the conversion value arrives net and after discount; a test purchase with the sale price and a voucher will show you. If your account does not deduct returns, use the expected contribution margin after returns for P × (1 − r) − k, see break-even ROAS. If you use contribution margin as your conversion value, break-even is 1 and the formula no longer applies.

Different margins need different targets, but not necessarily separate campaigns. Targets can also be set per ad group, which Google advises against. Separate campaigns are worthwhile when each one carries enough conversions. For Search and Shopping campaigns Google names at least 15 in 30 days; I would ask for considerably more. Otherwise you are left with a blended target: planned revenue divided by permitted ad spend. Suppose €60,000 at a target of 5.3 and €40,000 at a target of 3; together they give around 4.1. I would not start restructuring shortly before the peak; more on this in target ROAS by margin group. Whether a seasonality adjustment also makes sense depends on how reliably you can predict the jump in conversion rate.

The 30-day window works in both directions

A promotion before the promotion makes the Black Week discount smaller. Under Section 11 of the German Price Indication Ordinance (PAngV), every price reduction shown to consumers must state the lowest price of the last 30 days, and according to the CJEU a percentage discount refers to that price. For a start on 27 November, the window reaches back to 28 October. Worked example: normal price €100, an earlier promotion at €80 that has since ended, €70 on Black Friday. You can advertise 12.5% instead of 30%.

The same applies going forward. Until around 30 December, every Christmas promotion is measured against the Black Week price, so a promotion at €75 cannot be advertised as a reduction after €70. The depth of your discount in November is therefore also your price communication in December.

For the sale price annotation in Shopping, Google's own rule comes on top: in Germany the discount must be more than 5% and less than 90%, and the base price must have applied on 30 of the last 200 days. If you set the strikethrough price to €80 because of an earlier promotion, you lose the annotation until €80 has applied on 30 days. My conclusion: no price promotion in the 30 days before. How to set this up in the feed is covered in showing sale prices and strikethrough prices correctly. The legal reading is mine and applies to Germany; the wording of your advertising is a matter for your lawyer. For Austria and Switzerland, where prices are in CHF, check the legal position separately.

The price goes in the feed, the promotion gets reviewed in advance

The discount runs through sale_price with a date range and an explicit time zone, so that feed and shop switch in the same hour. A Merchant Center promotion is an add-on, and you do not have to bring its public start forward to get it reviewed. In a promotions data source, set promotion_effective_dates earlier than promotion_display_dates, with the same end date. The review then runs before the public start, and the approved promotion goes live when the display dates begin. The Promotion Builder lacks the start date attribute, so this only works through a data source.

During the review period, the promotion must be testable on the website and work as described. In my assessment it must not become a publicly usable price, because it would then count within the 30-day window.

November is not a budget reserve

At the end of the month a second limit applies alongside the daily one. After a budget change, spending for the rest of the month does not exceed the new daily budget multiplied by the remaining days, and in 2026 Cyber Monday falls on 30 November. On a single day a campaign can spend up to twice its daily budget, according to Google. So:

Daily budget ≥ the larger of peak ÷ 2 and planned spend from the day of change ÷ remaining days

Worked example: you plan €500 on Thursday, €2,000 on Friday, €1,000 each on Saturday and Sunday, and €2,000 on Monday. If you set €1,000 on Thursday, that just about covers the peak, but over five remaining days it only covers €5,000. Cyber Monday would be left with €500. What you need is €6,500 ÷ 5 = €1,300.

For a dedicated promotion campaign, the alternative is a campaign total budget: no daily limit, and the total is never exceeded. Google recommends at least 3 days for Search, Standard Shopping and PMax. Existing campaigns cannot be switched over, and the system spreads overspend and underspend across the remaining days. You therefore do not plan the Friday peak yourself.

Before every increase, check the target, not the current ROAS. Suppose a budget-limited campaign with a target of 2.5 is running at a ROAS of 5: in my assessment, with more budget it may slide towards 2.5. I would set such jumps shortly before the peak and reverse them afterwards, see winding back after Black Week.

The schedule for 2026

The dates are my suggestion for a promotion running from Friday to Monday; if you start earlier, move them forward. The Google details in this article are as of November 2026.

On the promotion day itself, leave the target alone, because in my assessment same-day conversions are still incomplete. Watch shop orders and stock instead. Bring sold-out products in the feed into line with the shop.