Blog5 min read

Black Friday ads cost 59 percent more. Buy them anyway.

In 22.6 million dollars of Meta spend, CPMs rose 59 percent in Black Friday week and cost per purchase fell 20 percent. The catch is which ads carry the week.

Somebody in finance opens the November dashboard and finds the CPM line. CPM is what you pay for a thousand impressions. It has gone nearly vertical. The message arrives before lunch. Can we pull back until prices calm down?

It is a reasonable question and the wrong one. Two datasets published this autumn point the same way. In the sale window, being seen costs more and a sale costs less.

What two datasets found

Lebesgue published the first on 16 October 2026. It covers 22.6 million dollars of Meta spend from September through November 2025, from the biggest spenders on its platform. For the median brand, in Black Friday week against September:

  • CPM rose 59 percent. On Black Friday itself it rose 84 percent, and on Cyber Monday 71 percent.
  • Cost per purchase fell 20 percent.
  • Conversion rate, counted as purchases per link click, rose 48 percent.
  • ROAS as Meta reports it rose 19 percent.

Confect published the second on 25 September 2026. It covers 2.6 billion impressions and 24 million dollars of spend on catalog ads, across 1,362 brands and retailers, for the whole of November 2025 against the rest of the year. CPM ran 45 percent higher, at 10.9 dollars against 7.5. Cost per purchase ran 10 percent lower. Conversion rate ran 25 percent higher.

The two do not match and should not. One measures a week against September. The other measures a month against eleven months, on one ad type. They agree on direction, which is the useful part.

CPM is the price of being seen. It was never the price of a sale.

It has already started

In Lebesgue's data the 2025 climb began in the week of 10 November, with a first daily rise of 13 percent. The year before it began in the middle of October. And it does not end with the sale. CPM was still 20 percent above the September baseline on 3 December.

So if your CPM moved this week, nothing broke. Take your baseline from September or October now, before the comparison gets muddy. Plan the budget through the first week of December, not through Cyber Monday.

The objection: these are vendors, and ROAS flatters a sale

Both points are right. Lebesgue sells analytics and Confect sells catalog ad tooling. Lebesgue says in its own write up that the data does not isolate one cause, and that its category groups are small. Confect calls its results correlations, not controlled tests. Neither states a country.

The purchases are also counted by Meta, in a week when people who were going to buy anyway buy on a discount. A 20 percent cheaper purchase does not help you if the discount took 40 percent of the margin. So do not take the lift on trust. Read the week on orders in your store and on margin after the discount. The claim that survives all of that is narrow. A rising CPM, on its own, is not a reason to stop.

The ad that carried September will not carry the week

Confect split its ads in two. Always on ads, and ads built for the month. The built ones cost more to show, 12.0 dollars against 10.6. They also returned more. Their ROAS lift was 37 percent, against 15 percent for the always on ads.

Biddyco looked at survival. In the ten largest US accounts in its data, from July to October 2026, only 34 percent of each account's top ten ads were still in the top ten two months later. Of the later top ten, 56 percent were brand new. The survivors held their click rate, and their CPM still rose about 20 percent as Meta moved budget to newer ads. Ten accounts is a small sample. It still matches what most of us see in an account.

Put the two together. Your best ad from September is probably not the one Meta wants to spend on in late November. And the ads that earn the most in the sale are the ones made for it.

What to do before the 27th

  • Judge the week on cost per purchase and margin after the discount. Leave CPM out of the decision.
  • Write down your September and October numbers today. That is your baseline.
  • Launch the ads built for the sale now, not in Black Friday week. Lebesgue's advice was to have them ready by late October. Late is still better than the day itself.
  • Keep fresh ads going into the account through 3 December. Costs stay up, and Meta keeps moving budget to what is new.

Let's talk about understanding your buyer,
and reaching them at scale.

Scale your business