Black Friday for DTC brands: the six-week ad plan.
Black Friday 2026 lands on 27 November. Cyber Monday is 30 November. Six weeks out is 16 October — and that date decides more about your Q4 than any creative you approve in November. By the time CPMs peak, the three things that determine the result are already fixed: how large your warm audience is, how many creatives you have in rotation, and whether your tracking can still count a purchase correctly at four times normal volume.
Most DTC brands treat BFCM as a four-day event with a two-week ramp. The brands that actually clear their Q4 number treat it as a six-week program in which the discount is the last thing that happens. Here is the plan, week by week.
The calendar that actually matters
| Window | Dates | Job | Share of Q4 budget |
|---|---|---|---|
| Weeks 6–5 | 16–29 Oct | Build the audience you will sell to | ~10% |
| Weeks 4–3 | 30 Oct–12 Nov | Produce and pre-test creative; lock the offer | ~15% |
| Week 2 | 13–19 Nov | Tease the offer, harvest the early-access list | ~10% |
| Week 1 | 20–26 Nov | Early access live to warm audiences only | ~20% |
| Sale | 27 Nov–1 Dec | Full spend, no experiments | ~35% |
| Week +1 | 2–8 Dec | Second purchase, non-buyer sweep, gifting | ~10% |
Weeks 6–5: buy the audience while it is still cheap
October traffic is the cheapest traffic you will buy all quarter, and it is the only traffic you can still convert at full price. Spend it on reach and list growth, not on last-click ROAS.
- Run broad prospecting to content and product pages, optimising for view content or add to cart rather than purchase. The point is pool size, not conversions.
- Grow the email and SMS list hard. An early-access signup collected in October costs a fraction of a November click and converts on Black Friday at several times the rate of cold traffic.
- Refresh every custom audience — 180-day site visitors, video viewers at 50%, engagers, past purchasers — and confirm they are actually populating.
- Check exclusions now. Recent purchasers included in prospecting is the most common way brands quietly overpay in the sale window.
Weeks 4–3: creative volume beats clever targeting
Targeting levers in 2026 are mostly gone; the auction rewards whoever feeds it more distinct creative. Plan for 20–30 shipped assets going into the sale, not four hero pieces. A workable split:
- ~60% offer-led — the discount stated plainly in the first second, in the first frame, in the thumbnail. BFCM shoppers are not being persuaded; they are scanning for the number.
- ~25% product demo — the thing working, in hand, unedited. This is what keeps CPMs down when offer fatigue sets in on day three.
- ~15% social proof — reviews, UGC, "sold out last year" scarcity that is actually true.
Produce in vertical 9:16 first and cut everything else from it. Pre-test the top five to seven variants in the first week of November at low budget on a purchase objective — you are not looking for winners yet, you are eliminating the ones that will waste sale-week spend. And test the offer too: a straight 30% off, a tiered spend threshold and a bundle will not perform the same, and the difference in average order value is usually larger than the difference in conversion rate.
Weeks 2–1: the pre-sale is where the margin is
Opening early access to your list and warm audiences on roughly 20 November does three useful things: it pulls revenue into a week where CPMs are 30–40% below the peak, it gives algorithms conversion data before the expensive days, and it tells you which creative and which offer to put the real money behind on the 27th.
Keep cold prospecting running through this window even if its ROAS looks unremarkable. Every cold visitor you buy on 21 November is a retargeting impression you get to serve on 28 November, when acquiring that person fresh costs roughly double.
What the auction does to your CPMs
| Period | Meta CPM index (Oct = 100) | Implication |
|---|---|---|
| 16–31 Oct | 100–110 | Cheapest audience-building of the quarter |
| 1–14 Nov | 115–130 | Still viable for prospecting; test here |
| 15–24 Nov | 140–165 | Warm audiences only start to make sense |
| 25 Nov–1 Dec | 170–220 | Peak. Spend, do not experiment |
| 2–15 Dec | 120–140 | Cheap again, and buyers are still buying |
These are the bands we typically see; your category and GEO will move them. The structural point does not change: a campaign that starts learning on 25 November is paying peak prices for education it could have bought in October at a third of the cost.
Tracking: the things that break at 4x volume
Every BFCM post-mortem we have run for a DTC brand has found measurement damage, not media damage. Fix these before 20 November, because you cannot fix them during the sale:
- Server-side events with proper deduplication. Browser-only pixels lose 15–30% of purchases in a normal week; under peak load and iOS defaults, more. Send the same
event_idfrom browser and server so nothing double-counts. - Pass real order value, after discount. Value-optimised campaigns bidding on pre-discount revenue will happily buy you unprofitable orders all weekend.
- Watch attribution windows. A 7-day-click window over a six-week program credits differently than 1-day-click. Pick one, note it, and compare like with like — not last year's dashboard against this year's setting.
- Load-test the checkout and the landing page. A 1-second delay at 4x traffic is a conversion-rate problem that will look exactly like an audience problem in your reporting.
The week after, where most brands stop
Discounted first orders are frequently break-even or worse once shipping and returns land. The profit in BFCM is the second purchase. In the first week of December, run three things: a non-buyer sweep to everyone who added to cart and did not check out, a cross-sell to weekend buyers at a lower discount, and a gifting angle to your whole warm pool while CPMs have fallen back. Then measure BFCM cohorts on 60-day repeat rate, not weekend ROAS — that number tells you whether the discount bought customers or bought orders.