]
Measure What the Platforms Can’t See.
Ad platforms can’t see the whole journey, so half your ads get graded wrong every week. This is how Northbeam reads BFCM in real time: what 2026 changed, the data points to keep open all weekend, and the thresholds that tell you when to act.
The Platforms Are Telling You the Wrong Story
“Over-reporting” is the wrong word for it. The real problem is that ad platforms can’t see all your touchpoints, so losing ads get labeled winners and winners get labeled losers.
Because each walled garden claims as much credit as it can and shares nothing with the others, in-platform reporting biases toward bottom-of-funnel campaigns that capture demand rather than the ads that created it.
C+DV launched October 2025 across Meta, TikTok, Snap, Pinterest, Axon, MNTN and Vibe.
The gap isn’t an error to reconcile; it’s the reason to measure independently. Study scale and results attributed to Northbeam (Data Partner).
“You’re scaling the wrong ad 50% of the time if you’re only using in-platform data.”
2026 Broke the Pattern
Before you set a single target, account for this: 2026 reversed the trends you’re comparing against. Meta ROAS declined and first-time CAC climbed through the first half of the year, erasing 12+ months of previous gains, which will quietly wreck naive year-over-year math. In Northbeam’s weekly benchmark (Jan 2025–Jun 2026), Meta ROAS peaked at $0.55 on BFCM week (Nov 30, 2025) and fell through H1 2026, while first-time CAC bottomed at $190.80 the week of Feb 1 and climbed.
Underneath that reversal, the mechanics changed too: clickthrough rates surged to highs while conversion rates sank to 12-month lows. More people are clicking and fewer are buying, so a CTR that looks like a win can mask a CVR that’s quietly collapsing.
Where Did You Land?
The spread across advertisers is wide. Median spend rose +12.38% year over year while median new-customer CAC rose +10.34% and median new-customer MER fell −5.73%. The top quartile still pulled well ahead of the bottom, so the useful question isn’t the average, it’s which side of it you were on.
Two cuts sharpen the benchmark. Higher-AOV brands weathered the year better, and MER and CAC worsened for advertisers of nearly every size, a reminder that “the market got harder” is not a personal failing but a baseline to plan against.
| Average Order Value | |||||||
|---|---|---|---|---|---|---|---|
| < $50 | $50–$100 | $100–$150 | $150–$200 | $200–$500 | $500+ | ||
| Spend | 25th | -21.5% | -16.3% | -16.6% | +2.7% | -13.7% | -0.6% |
| Median | +10.1% | +6.8% | +10.4% | +15.9% | +13.8% | +21.4% | |
| 75th | +54.0% | +42.3% | +33.0% | +52.0% | +27.5% | +51.5% | |
| Revenue | 25th | -22.6% | -17.9% | -15.4% | -11.2% | -3.8% | -2.3% |
| Median | +1.8% | -1.9% | +3.5% | +8.5% | +12.4% | +15.6% | |
| 75th | +32.1% | +29.4% | +31.3% | +36.9% | +25.0% | +56.0% | |
| Revenue (New) | 25th | -32.5% | -31.9% | -21.4% | -15.7% | -8.5% | -2.8% |
| Median | +6.3% | -7.3% | -1.9% | +5.4% | +10.7% | +15.6% | |
| 75th | +35.4% | +28.4% | +29.4% | +31.6% | +24.2% | +42.3% | |
| MER | 25th | -26.0% | -17.0% | -15.7% | -18.3% | -11.1% | -15.0% |
| Median | -1.4% | -0.1% | +1.3% | -5.3% | +1.4% | -2.0% | |
| 75th | +18.5% | +16.9% | +16.4% | +4.9% | +15.1% | +9.1% | |
| MER (New) | 25th | -36.3% | -25.9% | -20.0% | -22.7% | -13.9% | -17.7% |
| Median | -9.1% | -8.9% | -8.0% | -8.9% | +2.8% | -2.8% | |
| 75th | +10.5% | +5.2% | +10.3% | +8.0% | +18.4% | +13.7% | |
| CAC (New)↓ LOWER BETTER | 25th | +72.5% | +35.4% | +27.3% | +30.4% | +20.6% | +36.5% |
| Median | +11.4% | +11.8% | +9.2% | +12.3% | +7.2% | +13.3% | |
| 75th | -13.4% | -6.4% | -5.5% | +0.4% | -9.2% | -5.2% | |
| Company Revenue | |||||||
|---|---|---|---|---|---|---|---|
| < $5m | $5m–$10m | $10m–$20m | $20m–$50m | $50m–$100m | $100m+ | ||
| Spend | 25th | -26.9% | -7.5% | -13.6% | -6.2% | -6.9% | -4.2% |
| Median | +11.1% | +12.0% | +3.8% | +16.8% | +9.0% | +11.4% | |
| 75th | +53.1% | +28.2% | +35.3% | +43.4% | +29.0% | +30.4% | |
| Revenue | 25th | -30.8% | -16.2% | -8.1% | -7.0% | -11.1% | -8.6% |
| Median | +0.7% | -2.4% | +6.7% | +15.3% | +3.1% | +11.2% | |
| 75th | +43.6% | +19.5% | +28.3% | +34.8% | +20.4% | +28.8% | |
| Revenue (New) | 25th | -32.7% | -23.1% | -17.7% | -11.9% | -22.5% | -11.7% |
| Median | +1.6% | -5.3% | +8.0% | +13.0% | -2.0% | +6.4% | |
| 75th | +45.7% | +15.3% | +39.6% | +34.8% | +16.1% | +23.6% | |
| MER | 25th | -18.8% | -19.8% | -13.3% | -16.7% | -14.7% | -7.5% |
| Median | -1.3% | -7.2% | +6.1% | -0.6% | -6.1% | +3.4% | |
| 75th | +27.0% | +11.3% | +17.8% | +10.1% | +3.5% | +9.3% | |
| MER (New) | 25th | -25.2% | -29.2% | -15.8% | -21.0% | -23.7% | -10.9% |
| Median | -4.8% | -9.5% | +0.5% | -3.1% | -9.7% | -1.5% | |
| 75th | +29.6% | +2.9% | +17.6% | +8.8% | -0.1% | +6.5% | |
| CAC (New)↓ LOWER BETTER | 25th | +43.8% | +52.0% | +24.6% | +22.2% | +32.4% | +18.7% |
| Median | +8.7% | +17.8% | +9.5% | +9.5% | +12.9% | +6.3% | |
| 75th | -24.5% | +4.1% | -10.7% | -4.8% | +3.7% | -4.1% | |
Two Lenses That Simplify BFCM
The data you focus on is only as good as the strategy for reading it. Lets distill it into two ideas. The first is conversion lag: how long an ad runs before it hits peak performance. A swimwear brand might see a 1.0x same-day Facebook ROAS grow to 1.35x after 30 days, a 35% lift. Lag grows as BFCM approaches, because shoppers already know the sale is coming, so the same-day ROAS you can tolerate should fall as you near Black Friday. The job is to set a healthy one-day number you can act on today, knowing it will grow.
The second is flighting analysis: for each day, divide that day’s share of revenue by its share of spend. Above 100% and the day’s dollars punched above their weight (you probably underspent); below 100% and you may have overspent. Overlay when your offers turned on, and last year’s customer behavior tells you how to plan this year. The pattern it reveals is consistent: Black Friday is where everyone overspends, and the quiet Saturday and Sunday are where efficiency actually lives.
What to Watch
This is the board an operator keeps open from Thanksgiving through Cyber Monday: eight metrics in two tiers. Tier 1 is pacing, blended MER on cash accounting, spend charted against revenue on one axis so you see divergence instead of computing it, and revenue split total-versus-first-time (a promo pulls returning buyers forward, so a healthy blended number can hide an acquisition collapse).
Tier 2 is decisionmaking, anchored on the metrics Northbeam treats as most important at all times, new-customer ROAS and CAC, plus New Customer % and the underrated 1-day %, the guardrail that tells you whether you’re being paid for demand you’re buying now or demand you already generated. Deltas are day-part matched: at 2pm you compare accumulated-to-2pm against accumulated-to-2pm yesterday.
The board an operator keeps open all weekend. Example read; colors follow the threshold rules below. Attributed to Northbeam (Data Partner).
Thresholds That Trigger Action
Not every business measures the same, so these are the reads to watch to avoid the serious pitfalls.
| Keep going | Investigate | Change now | |
|---|---|---|---|
| Blended MERcash | At or above pre-BFCM baseline | Below baseline for 2+ day-part-matched reads | Sustained below baseline while spend climbs. Cap channels before adding |
| Spend pacingvs plan | Within plan; flighting ratio ≥ 100% | Ratio 90–100% on a day you’re scaling into | Ratio well below 100% two days running |
| nROASaccrual · CO · 1d | At or above lag-adjusted 1-day target | Within ~10% below target | Clearly below target at channel level |
| New customer CAC | At or below channel target | Rising while blended CAC holds flat | Rising while New customer % also falls. You’re buying existing customers |
| New customer % | Holding at or above profitable-day baseline | Drifting down while blended ROAS looks fine | Falling materially. Audit exclusions and optimization events |
| 1-day % | Within ~5 pts of last year’s benchmark | 10+ pts below benchmark | ~half of new revenue from outside the 1-day window. Stop scaling, audit before day two |
| Email / SMS contribution | At or above last year’s same-window share | A few points below | Considerably lower than prior years. Send more emails now |
The decision key for the board above, verbatim from Northbeam.
What You Can Change Mid-Weekend, and What’s Locked
Once the weekend is live, the fastest lever is also the safest, and the tempting levers are the ones to leave alone. Know your goal before you touch anything: it isn’t to make as much revenue as possible, it’s to acquire as many new customers profitably as possible.
- SMS & email frequency. When in doubt, send more. Zero learning-phase risk, immediate effect, measurable against last year’s benchmark within hours, your break-glass tool.
- Channel budgets, within profitable thresholds. If something acquires customers profitably, crank it. But moving spend between channels is harder than you think, focus on one or two, and don’t test anything new on the big weekend.
- Your attribution model. Non-negotiable. Don’t move the goalposts on the big day, you set the goals in advance for a reason.
- Landing pages & site structure. Promo pages work because they’re tightly matched to the offer; they’re built long before. Shift them in real time and you’ll only lose conversion rate.
What Operators Get Wrong
Four mistakes show up again and again when brands read their own BFCM data.
| The mistake | The correction | |
|---|---|---|
| Same-day judgment | Judging pre-BFCM spend on same-day return, early-November dollars look like failure on day-one ROAS. | Those dollars fill the funnel and the return lands during the sale. This is the whole reason flighting analysis exists. |
| Reconciling the gap | Reading last-click and treating the platform-vs-Northbeam gap as an error to fix. | The gap is the point: platforms each claim the full order; Northbeam divides one order across the journey. It won’t line up, by design. |
| Headline revenue | Scaling on headline revenue without checking the 1-day %. | Great revenue with a collapsing 1-day share means you’re being paid for demand you already generated, adding budget buys nothing. |
| The biggest day | Over-weighting Black Friday, the most expensive auction and the day efficiency reliably goes backward, then failing to cut fast enough. | Pre-commit cut triggers, cut in the morning rather than at 5pm, and plan Saturday and Sunday down rather than flat. |
The four most common BFCM data-reading mistakes, and how to correct them. Attributed to Northbeam (Data Partner).
This chapter’s benchmark data, frameworks and thresholds were supplied by Northbeam.