How to Build a BFCM Creative Batching Plan

· Creative

How to Build a BFCM Creative Batching Plan

Most brands we audit start briefing Black Friday creative in the first week of November. The assets land in the account at the exact moment spend starts climbing, which leaves no time to learn which concepts work before the budget multiplies.

The multiple is bigger than most teams plan for. We pulled last season's spend through Data Druid, our in-house account intelligence platform that sits on top of every ad account we manage. Across the 12 accounts with enough history to measure, the median account spent 3.8x a normal day on its peak BFCM day. A quarter of them went past 9.9x. Over the heaviest 30 consecutive days of the season, the median account still ran at 1.8x its normal spend.

Spend scales like that while the creative library usually stays flat. Frequency climbs with every extra dollar, pulling CPMr up alongside it. By Cyber Monday the account is paying peak auction prices to show the same handful of ads to the same people.

This is the batching plan we use to get 30+ variations built in October, tested before the offer goes live, and rotated through the season without a second production budget.

4 more mistakes we see in every BFCM account

Late briefing is the first problem. These 4 follow close behind it.

Volume without variety. We regularly see BFCM batches built from 1 hero image with 30 headline, color and discount swaps. Meta's Andromeda system, which decides which ads are eligible to enter each auction, appears to group near-identical creative together. In practice those batches behave like a single ad served to a single audience. The account ends up with 30 live assets and almost no movement in reach.

[embed:https://x.com/PhilKiel/status/2026310417103126532]

Every asset leads with the offer. %-off statics do a real job during BFCM. When every asset in the account leads with the discount, though, nothing is doing acquisition work. Budget ends up cycling through people who were already going to buy. In the numbers, that shows up as strong platform ROAS sitting next to a rising CPMr and a falling new visitor share.

Rebuilding creative for every offer phase. Early access, Black Friday weekend and Cyber Monday often carry different messaging. Teams that don't plan for it end up re-briefing from scratch 3 times in 4 weeks.

Letting exclusions slip. Every early access purchaser stays in your prospecting pool through Cyber Monday unless the exclusion lists get refreshed. You keep paying to reach people who already bought. At several times normal spend, that wasted frequency compounds fast.

Build 30 ads from 6 concepts

The plan works because the 30 variations come from 6 concepts. The concept is the only layer that costs real money. A concept means a different angle, person, problem or setting, which changes who Meta shows the ad to. Format and offer sit on top as layers that cost editing time only.

From 6 concepts, the batch breaks down like this.

  • 6 vertical videos, 1 per concept, shot once with the final few seconds left open for the offer.

  • 6 motion assets, 1 per concept, built from the video footage and static layouts.

  • 18 statics, 3 per concept, 1 for each offer phase across early access, Black Friday weekend and Cyber Monday.

That adds up to 30 assets with all of the production spend sitting in the 6 concepts.

When picking the 6, mix the jobs. Building 6 versions of your best seller hands Meta the same ad 6 times. In most accounts we want 2 acquisition concepts that skip the discount entirely, such as a Yapper (1 person talking straight to camera) or a Partnership Ad with a creator. Add 2 proof concepts like a product demo or a review-led piece, plus 2 offer-forward concepts that handle the conversion work. The acquisition concepts keep CPMr in check once spend multiplies because they reach people who haven't seen the brand yet.

Shoot once and cut everything from the footage

Production order matters more than production budget. Video comes first, with every other format derived from it and nothing getting its own shoot.

Plan 1 shoot day to capture all 6 videos, or 1 creator batch if you're running Partnership Ads. Yapper concepts cost almost nothing to produce because the format needs a single person in a single location delivering a single take. You can film several in a morning by moving between rooms instead of booking new sets.

[embed:https://x.com/PhilKiel/status/2028137950173864103]

Stills pulled from that footage become the base for the statics. That keeps the visual language consistent across formats while removing the need for a separate photo shoot. Designers then build the motion assets from the static layouts with short video cutaways dropped in, so they're assembling instead of starting from a blank page.

The offer always lives on its own layer, as an end card on video and motion and as a text band on statics. When messaging moves from early access to Black Friday to Cyber Monday, you swap a template across 30 assets in an afternoon instead of writing a new brief.

Test the concepts in October while the auction is quiet

Black Friday lands on November 27 this year, with Cyber Monday on November 30. That leaves about 8 weeks from the start of October. Here is the calendar we work back from.

  1. In the first 2 weeks of October, lock the 6 concepts and brief them. Include the end card and text band specs so the offer layer is designed in from the start.

  2. In mid-October, shoot everything in 1 block. Cut the 6 vertical videos first.

  3. From the last week of October into early November, run all 6 concepts with evergreen messaging and no discount. Read thumbstop rate, CPA, CPMr and new visitor share for each.

  4. By mid-November, take the 4 strongest concepts into the BFCM build and finish their derivatives. Hold the other 2 back fully built as a reserve.

  5. From around November 20, launch early access with the phase 1 offer layer on the 4 live concepts.

  6. Over Black Friday weekend, swap to the phase 2 layer across the same concepts. Launching anything new at this point drops it into a learning phase at peak prices.

  7. For Cyber Monday, swap to phase 3. Bring in the reserve concepts if frequency is climbing.

The October test is the step most accounts skip. Auction pressure sits at normal levels during October, which makes it the most efficient window in the quarter to buy learning. It also means the concepts carrying peak-week budget have already proven they can find new people.

How to tell when the batch needs rotating

CPMr, the framework our Director of Paid Social Phil Kiel popularized, stands for Cost Per Mille Reached. It measures the cost to reach 1,000 unique people, calculated as CPM multiplied by frequency. CPMr rises during BFCM in almost every account, so the headline number on its own tells you very little. The useful read is which of its 2 inputs is doing the climbing.

If CPM is rising while frequency stays flat, you're looking at auction pressure from every other brand bidding into the same week, which fresh creative won't fix. Frequency rising alongside it means the budget is cycling through the same people. That's the signal to bring in the reserve concepts and confirm the purchaser exclusions have been refreshed since early access.

[embed:https://x.com/PhilKiel/status/2029701498268176617]

In a normal month we read CPMr week over week. During peak week, check it daily against your own October baseline. At several times normal spend, a saturated ad set can burn through a week's budget before the weekly report catches it.

If you're reading this in the first week of October, you still have time to brief 6 concepts this week and get them testing before the auction gets expensive. The brands that come out of BFCM with margins intact are the ones that decided which ads would carry peak week a month before it started. For the full breakdown on calculating and reading CPMr, seePhil Kiel's CPMr guide.


Methodology: Spend multiples come from 12 anonymized Taikun Digital accounts in Data Druid with sufficient history, comparing peak-day spend and the heaviest 30 consecutive days between October 15 and December 20, 2025 against an August 2025 baseline.


Back to all articles