Every year, brands come into Q4 thinking they need to reinvent their creative strategy.
They brainstorm new concepts, brief fresh ideas, and hope something sticks. But BFCM is the worst possible time to test. CPMs are climbing, the feed is saturated, and every pound spent on an unproven idea is a pound not spent on something you already know works.
The good news is that your best creative for peak probably already exists in your account. It just needs finding, reskinning, and deploying with the right splits.
And the timing matters more than anything else in this chapter. You can change an offer the day before you go live. You can build a landing page in a short window. You can't produce UGC in a condensed window. Ideally, your Black Friday creative is done by the end of October, which leaves November free for iteration and for preparing the gifting and Christmas creative you'll need the moment the sale ends.
Everything in this chapter comes from auditing the £32M we spent on Meta last November: what converted, what fatigued, and which old winners came back strongest with a fresh skin.
We break it all down in this episode of D2C Diaries, from audience building in October to funnel splits and the creative that carried peak week:
https://www.youtube.com/watch?v=nDpF5MRm2E0
A proper audit looks at two periods, because each one tells you something the other can't.
Last year's BFCM period shows you how your audience behaves in Q4 specifically. That can be very different from the rest of the year. For some businesses, gifting becomes as much as half of revenue during peak. That shift only shows up in last year's data, and it changes everything from creator choice to messaging, because the person buying a gift looks nothing like the person buying for themselves.
Your last 90–180 days show you what's efficient right now. That means which personas are driving the best CPAs, which angles are landing, and which formats the algorithm is currently rewarding. Last year tells you how the period behaves. Recent data tells you what's working today.
Break down every offer you ran last year and, more importantly, every way you framed it. Compare them on CPA, AOV, and spend. As we covered in the offer chapter, the same offer framed two different ways can deliver close to double the ROAS.
You don't need to do this manually. By connecting Meta's API, you can automate a report that cross-compares every offer and angle across your creative. You can even build it as a Claude skill and rerun it whenever you need to. The output should tell you quickly which framing was most efficient, which drove the highest AOV, and which the algorithm was most willing to spend behind.
Look at where your spend and sales went by age and gender, then compare that with who actually appears in your creative. The gap is often bigger than you'd expect.
In one account we audited, most sales came from women aged 35–54. But the people on screen in the creative were mostly men or women in their mid-twenties. The biggest opportunity sat with a demographic the creative was barely showing.
This matters more now than ever. With Andromeda, Meta relies heavily on the signals inside your creative to decide who to serve it to, and the person on screen is one of the strongest signals you have. If your casting doesn't match your buyers, you're making the algorithm's job harder.