Every source agrees: Meta is getting dearer heading into Black Friday
Northbeam's Meta CPM is running about 24% above last year, Meta's own price per ad rose 14% in Q3, and Common Thread's brands have seen Meta CPMs jump 17–43% from October to November in every year since 2022. Three in four of Foxwell's 600 buyers name rising CPMs as their top Q4 worry. The twist: Northbeam's Meta ROAS is up about 30% and new-customer CAC down a third, so efficiency is holding for now.
“A generational Black Friday Cyber Monday, and not in a good way.”
Weekly channel tracker
Forty weeks of Northbeam's weekly platform tables, from w/e 28 Dec 2025 to w/e 18 Oct 2026. Pick a metric to see every platform side by side, or pick a platform to see all its metrics.
Cumulative change compounds Northbeam's weekly moves from the first week in the range, so every platform starts at 0% and you can see which one has drifted most. Chained figures drift over long periods because each week compares a slightly different set of accounts; use them for direction and turning points.
Bars are the week-over-week % change Northbeam published. Green is better for advertisers (lower costs; higher CTR, CVR, ROAS), red is worse. The figure in each corner is the cumulative change over the range shown. Weeks Northbeam skipped are left blank.
Spend is up, revenue lags, new customers cost more
Northbeam's monthly, half-year and quarterly year-over-year cuts. All figures compare advertisers present in both periods.
Creative volume is the lever: winners ship two to three times more ads
Everything Northbeam has published on creative output, testing and measurement in the past year: ads launched by spend level, creative volume by sector, how in-platform data misjudges new ads, and the UGC and optimisation playbooks behind the top accounts.
Each line is a monthly ad-spend band. "vs band's own average" shows each month as a % of that band's 2025 monthly average, so seasonality is comparable across bands: every band peaks in November.
Mean vs median live Meta ads per brand. A mean far above the median means a few very high-volume brands in that sector (Pets: mean 1,382, median 357).
Last season's Black Friday, set against this one
BFCM 2025, the post-BFCM come-down and this autumn's numbers side by side. Projections are scenarios built from 2025's seasonal shape and are labelled as such.
Meta CPMs are running about 25% above last year, and the gap is widening into Q4
Northbeam's weekly Meta CPM across every account it tracks since January 2025, its monthly CPC, CTR, conversion, ROAS and CAC charts, the weekly index for Meta, and the Q3 2026 benchmark heatmap. Values read off Northbeam's charts are marked ≈; its labelled peaks are exact.
Q3 2026 vs Q3 2025 is the like-for-like read on the quarter just closed.
Solid lines are the weekly series from Northbeam's June charts. Dashed lines continue with monthly values read from its October posts, which use a slightly different cut (BFCM CAC $309 there vs about $270 here), so read the two as direction rather than one series.
Northbeam's headline: ad spend up more than 30% on average in Q3. Accounts spending $200k–$500k a month did best at the median (revenue +8.1%, new-customer revenue +6.3%); the $500k–$1m band saw median new-customer CAC rise 13%.
Research library
Every Northbeam post with data in it from the last 12 months, newest first, with the key numbers and the datasets transcribed from its charts.
- Northbeam's samples change between posts (all accounts, accounts present in both periods, one-day clicks, accrual or cash). January and February 2026 appear twice with different values for this reason.
- Weekly index weeks run Saturday–Friday until mid-April 2026 and Sunday–Sunday after. Not published: 29 Dec – 2 Jan, 31 Jan – 6 Feb, 14 – 21 Jun 2026. No weekly index ran Oct – mid-Dec 2025, so BFCM 2025 comes from the Cyber Week reports.
- The week of 7–14 Jun 2026 was published twice (18 and 25 Jun); the later version is used.
- Snapchat CPC shows +12.25% in four consecutive weeks (28 Feb – 27 Mar 2026), likely a carry-over error in Northbeam's graphic.
- Until w/e 10 May 2026, budget share excludes "All other platforms"; the share desk rebases every week to the eight core platforms so the year is comparable.
- Meta's average first-time CAC fell 33% YoY in Q3 while the median account's new CAC rose in most spend bands. Benchmark against your own spend band.
The Cyber Week desk
Last year's day-by-day results are set as the baseline. From Tuesday 24 November the scheduled update logs each new figure from Northbeam, Triple Whale and the daily trackers here, day by day against 2025.
Bars: Northbeam's BFCM 2025 results, YoY vs 2024, by day. Bright markers: BFCM 2026 vs 2025 as each figure is published. The two years use different comparison bases, so read the 2026 markers as "is this year running hotter or cooler than last year's growth?"
Meta's price per ad is rising fastest in the US, at +23% a year
Each quarter Meta reports how many ad impressions it served and how much the average ad cost, both year on year. It is a market-wide read on auction inflation, covering every advertiser on Meta, to set against Northbeam's DTC sample.
Lines are YoY % changes Meta reports in each release. When impressions grow fast (new placements such as Threads and Reels, and growth in Asia-Pacific), the average price is held down. When impression growth slows, price takes over: the second half of 2026 is that pattern.
Roughly one ad in twenty becomes a winner. Volume is how you find it
Motion analysed 550,000 Meta ads from 6,000 advertisers ($1.3 billion of spend, September 2025 to early January 2026). A winner is an ad that spends at least ten times its account's median ad. Motion rebranded as Runneth in 2026; the Thumbstop newsletter and reports carry on under the new name.
The most-saved ads, overall and in 31 industries
Runneth ranks the top 30 Meta, TikTok and Instagram ads by how often its users save them, on a main list plus 31 industry lists (found through a hidden sitemap, all marked "weekly"). Past weeks come from the Internet Archive; from 28 October the Thursday update captures all 32 lists.
What Motion publishes, and how often
Motion's sitemaps (runneth.com and motionapp.com) list 300+ pages but no newsletter archive. Thumbstop goes out every Sunday by email only; its long-form pieces reappear as "Thumbstop" posts on the blog, and its data lands in quarterly "Thumbstop Pulse" reports.
What actually moves sales: Haus's lift tests, channel by channel
Haus runs geo holdout experiments for its customers and publishes what thousands of them show. Long research reports arrive every one to three months; between them, CMO Olivia Kory and @HausAnalytics post smaller data snapshots on X several times a month. Both are gathered here.
Where each industry stands this month
A weekly snapshot of Triple Whale's free benchmarks: brands with $1–10M annual revenue, trailing 30 days, 12 DTC industries, every channel. Only real snapshots are kept. Triple Whale's tool will return past date ranges, but those are a scaled seasonal index rather than weekly data, so history builds from 28 October 2026.
Bars run from the 25th to the 75th percentile brand; the marker is the median. Channels with too few brands in an industry are left out by Triple Whale.
The long view
Weekly, monthly and quarterly figures from Northbeam, Meta, Common Thread, Triple Whale and Haus, plotted on the same calendar so you can see where they agree. Each publisher measures a different set of brands in a different way, so compare direction and turning points, not exact levels. Values marked ≈ were read off a published chart.
Lines: Northbeam's weekly share of core platform spend. Markers are other publishers' quarterly snapshots of the same platform: ▲ Northbeam share of all tracked spend, ■ Common Thread portfolio (Meta and Google include all campaign types), ◆ Haus share of wallet. Their bases differ, so markers sit above or below the line by design.
Left axis: the DTC Consumer Confidence Index (Common Thread × KnoCommerce, 5,000 shoppers a day; 100 = neutral). Right axis: Northbeam's median brand revenue change vs last year, and Meta's ad impressions change vs last year from its earnings.
What 600 media buyers and agency owners are saying
Foxwell runs Foxwell Founders, a paid community of agency owners and brand marketers managing over $1B a month in ad spend. It doesn't publish a running data series: its numbers come from member surveys (a yearly agency benchmark and, this year, a Q4 outlook survey with Northbeam) and from field reports in its blog, which posts two or three times a week. Treat this desk as the practitioner view beside the platform data.
Only 28% look at contribution margin daily, against 72% for ROAS. 43% still pace spend from a manual spreadsheet.
"No difference" is stated as "a third" in the write-up.
What 299 brands spend, and what each channel is really worth
Common Thread Collective runs growth for hundreds of DTC brands and publishes their pooled data: the DTC Index (200–300 stores with two or more years of history), interactive research pages, and a daily consumer confidence survey with KnoCommerce. Their weekly DTC Index email has no public archive, so this section is built from the published benchmarks, podcasts and blog.
Bigger spenders diversify: above $500K a quarter, TikTok, AppLovin and YouTube take a combined 4–5% of spend. Tick the box to zoom in on the smaller channels.
Other feeds worth bringing into the newsroom
Sources that publish data on a regular cadence, grouped by what they would add.