Meta CPMs follow a predictable annual shape. They climb from late October, peak during Black Friday week and the first three weeks of December, then fall sharply between December 26 and mid-January.
Here is what third-party datasets report, plotted as bands rather than as a line, because the spread between sources is wide enough that a point estimate would be dishonest.
| Period | Reported CPM movement versus baseline |
|---|---|
| Q4 overall | Up 20 to 50 percent |
| Black Friday week | Up 50 to 80 percent, with some categories at 2 to 3 times normal |
| Key e-commerce categories at peak | Up as much as 66 percent |
| December 26 to January 15 | Down 40 to 60 percent |
One source reports Q4 2025 US CPMs averaging $25.49, roughly 22 percent above Q1 2025.
Treat every number above as a reported range from an outside organisation rather than a forecast for your account. Your CPM is set by your auction, your audience, your objective and your creative, and the spread between verticals is larger than the seasonal effect in many accounts.
The planning implication is not spend less in December. It is that the same budget buys materially different volumes of impressions depending on when you deploy it — and that the post-Christmas window is the cheapest inventory of the year while most advertisers have stopped paying attention.
Why the curve exists
The curve is a demand artefact, not a pricing decision. The Meta ads auction allocates impressions to the highest total value bid, and inventory is close to fixed in the short run. When more advertisers compete for the same finite impressions with higher bids, clearing prices rise.
That framing has a consequence people miss. The December CPM increase is not a penalty applied to you. It is evidence that other advertisers concluded those impressions were worth more. Whether they are worth more to you depends entirely on your own Q4 conversion-rate and average-order-value lift, which is a number from your store rather than from a benchmark table.
Converting the curve into a budget decision
The question that actually matters is not will CPM rise. It is does my conversion value rise faster than my CPM.
Define seasonal margin headroom as:
Headroom = (revenue per 1,000 impressions) ÷ (CPM)
Compare that ratio in the peak period against the same ratio in your baseline period. If the peak ratio is higher, scaling into the expensive window is correct even though the CPM is worse. If it is lower, you are buying volume at a loss and calling it seasonality.
Worked example. Model inputs chosen to demonstrate the calculation. Not AdRiseLab customer data and not a benchmark.
| Input | Baseline (modelled) | Peak week (modelled) |
|---|---|---|
| CPM | $18.00 | $30.60 |
| Click-through rate | 1.1% | 1.3% |
| Clicks per 1,000 impressions | 11.0 | 13.0 |
| Conversion rate | 2.4% | 3.6% |
| Orders per 1,000 impressions | 0.264 | 0.468 |
| Average order value | $62 | $71 |
| Revenue per 1,000 impressions | $16.37 | $33.23 |
| Revenue ÷ CPM | 0.91 | 1.09 |
In this model the CPM rose 70 percent and the ratio still improved, because conversion rate and order value rose faster. Scaling is justified.
Now change one input. Hold conversion rate at 2.4 percent instead of 3.6 percent, and leave everything else identical: revenue per 1,000 impressions falls to $22.15, the ratio falls to 0.72, and the same CPM increase now destroys margin. The seasonal decision is decided by your conversion lift, not by the CPM headline.
Run the arithmetic on your own inputs before committing Q4 budget. The Meta ads ROI calculator and the industry benchmark tables are the starting points, and your order ledger supplies the conversion rate and average order value that actually decide it.
The four dates that matter more than the month
Late October. Costs begin drifting up. This is when creative testing is still cheap and still statistically useful. Testing in the first week of December is testing at peak cost on a compressed timeline, which is the most expensive way to learn anything.
The week before Thanksgiving. CPMs step up ahead of the event rather than during it. Budget increases applied on Black Friday morning arrive after the price has already moved.
The first three weeks of December. Sustained peak. Shipping-cutoff messaging and gifting angles carry this window. Delivery is volatile, so do not read a single day as a trend.
December 26 to January 15. Reported CPM declines of 40 to 60 percent as large advertisers pull back. This is the highest-value window in the calendar for three activities: prospecting at low cost, gift-card and post-holiday offers, and creative testing for Q1 at a fraction of December's price. Most accounts are dark. That is the opportunity.
For the week-by-week version of the peak itself, see the BFCM 28-day launch plan.
The learning-phase trap
A budget increase large enough to re-enter the learning phase during peak week is expensive twice: once because impressions cost more, and once because the delivery system is re-learning while they do.
Two practical guardrails:
- 1.Raise budgets before the cost step-up, not during it. The learning period then happens on cheaper inventory instead of on the most expensive impressions of the year.
- 2.Prefer adding budget at the campaign level over restructuring ad sets. Structural edits reset more than budget edits do.
See how to exit the learning phase faster and scaling without losing ROAS for the mechanics underneath both.
One thing that changed for Q4 2026
Placement selection is disappearing from ad sets, replaced by value rules that suppress a placement rather than switch it off. That matters here for a narrow but expensive reason: a Q4 performance comparison that spans the rollout mixes two different delivery regimes, and a placement mix that drifts during peak week will move your CPM independently of the season. Annotate the date you observed the change in your own account before you read any year-over-year Q4 chart. The detail is in what replaced placement controls.
What this article is not claiming
It is not claiming your CPM will rise by a specific percentage. Reported ranges span 20 to 80 percent depending on source, week and vertical, which is wide enough that a point estimate would be dishonest.
It is not claiming the post-Christmas window is cheap for everyone. It is cheap because demand falls, and in some categories consumer demand falls with it. Gift-driven categories and habit-driven categories behave very differently in January.
It is not claiming these ranges measure causality. They describe observed price levels across advertiser populations, and the composition of those populations changes seasonally too — which is part of why the ranges are so wide.
AdRiseLab generates image and video ad creative for Facebook and Instagram from a connected Meta ad account and reports revenue against the campaigns that produced it, separating verified orders from browser-reported ones. Budget, bidding and campaign structure stay with the advertiser, which is the half of this decision no tool should be making for you.
Related Reading
Why Meta CPMs jumped in 2026 covers the structural cost increase underneath the seasonal one, and the industry benchmark tables give the per-vertical context these ranges sit inside. For the peak week itself, use the BFCM 28-day launch plan. For the cheap testing window afterwards, the creative testing budget calculator shows how much less the same test costs when CPMs fall.