If you are building your Q4 plan off what happened last Q4, you are about to underprice your media.
Here is the trap. Most of the reporting on holiday 2025 said the same thing: Meta CPMs were flat, even down. That was true. It was also a one-time gift that has already been taken back, and the brands that budget as though it will repeat are going to spend the first three weeks of November wondering why their numbers look nothing like last year's.
Tinuiti's retail panel found Meta CPMs down 13% year over year in the seven days before Thanksgiving, and roughly flat across the Cyber weekend itself. Across the full quarter, Meta CPMs came in 7% below Q4 2024, with Facebook specifically down 13%.
That was not softening demand. It was supply. Meta pushed an enormous amount of new Reels inventory into the auction, and impressions grew about three times faster than price. Instagram Feed fell to 27% of impressions, an all-time low at the time (it has since slipped to 26%) — while Stories took 43% and Reels 29%. When a platform floods the auction with cheap placements, average CPM falls even as competition for the good placements rises.
Which is exactly what happened underneath the average.
Triple Whale's data on 33,000 DTC shops — $607M in ad spend over BFCM — shows those brands paid a Meta CPM of $22.26, up roughly 8% year over year. Big retail paid less. DTC paid more. Same platform, same weekend, opposite direction.
The gap between those two panels is the whole story. Large advertisers absorbed the cheap Reels inventory. Smaller DTC brands, concentrated in Feed, bid against each other for placements that were getting scarcer. Triple Whale measured Reels CPMs running 30–40% below Feed over the same period — the discount was sitting right there, and most DTC brands did not take it.
Meta reports its average price per ad every quarter. That figure is global and blended, so it is not your CPM — but as a trend line it is unambiguous:
Ad price growth doubled and stayed doubled. Meanwhile impression growth decelerated from 19% to 14%. The cheap-inventory tailwind that suppressed last year's holiday CPMs has largely been absorbed, and this year's seasonal premium is going to stack on top of an already elevated base rather than a discounted one.
Tinuiti's Q2 2026 read puts a finer point on it: Meta CPMs up 10% year over year, the largest increase since Q1 2022, on just 2% impression growth.
Last year the auction got cheaper while you weren't looking. This year it is getting more expensive while everyone is looking.
eMarketer's July 2026 forecast has US holiday retail growing 4.1% — the second consecutive year holiday growth trails the full-year average. So media prices are climbing at roughly triple the rate of the demand they are chasing. That compression is arithmetic, not strategy, and no amount of bid tuning fixes it.
Two more things are different this year, and both are creative problems rather than media problems:
The second one has a real creative implication. If a meaningful share of shoppers is asking an assistant to compare three products before they buy, vague claims stop working. Specifics, real reviews, and clear differentiation get summarized well. "The best hoodie ever" does not.
Black Friday 2026 lands on November 27, leaving 28 days to Christmas — one more than last year, and a more forgiving gifting runway than 2024 gave anyone. That is the good news, and it is most of the good news in this section.
Because what changed last year was not the calendar. It was where inside the calendar the money moved. Triple Whale found DTC brands already spending $396.6M in the four days before BFCM — 65% of what they spent on the BFCM weekend itself. And the share of BFCM revenue coming from new customers fell from 59.7% to 48%. If you are still treating the four-day weekend as your acquisition moment, you are arriving after the auction has been bid up and after the new customers have already been bought.
Jonathan Snow's BFCM 2025 post-mortem across his client portfolio found every paid media efficiency metric — ROAS, CAC, CPP — deteriorated across every platform, even as those brands increased spend more than 50%. His standout case was Free People, which peaked November 14–15 on an "Everything Sale" and saw minimal incremental lift from its Cyber Monday 40% promo. Going earlier beat going harder.
Meta's own 2026 holiday guidance splits the season into four phases — discovery in October, deals over BFCM, gifting in December, fresh start in January. The useful part is the reasoning Meta gives for the October start: Advantage+ campaigns need six to eight weeks of learning before peak. Common Thread Collective puts the practical deadline bluntly — brands not live by mid-October miss the window.
Working backward from November 27:
If your Q4 plan assumes last year's CPMs, add a working assumption of 10–15% higher base cost before any seasonal premium. Historically that premium has run 17–43% over baseline. Nobody has published a credible Q4 2026 CPM forecast — anyone who gives you one made it up — but the direction is not ambiguous.
Q4 2025 was cheap for large advertisers and quietly expensive for DTC. Q4 2026 will be expensive for everyone, into decelerating demand, on the shortest gifting calendar available, with discounting no longer doing the persuasive work it used to.
The lever that still moves is creative — volume of it, tested early, built for the placements that are actually cheap. Everything else is a rounding error on a rising CPM.
If you need assets in market by mid-October, the production window is now. Take a look at our work or the Content Bank package, which is built for exactly this — a full creative arsenal in one production cycle rather than a scramble in November.
Sources: Meta Q4 2025, Q1 2026 and Q2 2026 investor results; Tinuiti Black Friday/Cyber Monday recap (Dec 2025) and Digital Ads Benchmark Report Q4 2025 & Q2 2026; Triple Whale BFCM 2025 benchmarks (33,000 shops, published Aug 2026); Adobe Analytics 2025 holiday recap (Jan 2026); eMarketer holiday forecast (Jul 2026); Zeta Global 2026 Holiday Shopping Trends; Meta 2026 Holiday Insights Center via Social Media Today (Aug 2026); Common Thread Collective (Aug 2026); Jonathan Snow BFCM 2025 post-mortem (Dec 2025).
Assets need to be in market by mid-October. The Content Bank delivers 20 in one production cycle.