Creative Machine. That’s the answer to 2025 BFCM Creative Playbook.
What does that mean?
Why now? Meta Andromeda Update
Meta’s next‑generation ad retrieval engine requires to feed the AI many high‑quality, diverse concepts. If your creatives or data don’t make it through retrieval, they’ll never hit the auction.
Recent debate between Jordan M and Charley T which went viral on Twitter also confirmed Andromeda pushes for high volume, highly diverse creative volume.
On creative diversification:
- AI ignores near‑duplicates. Focus on new angles (persona, pain point, benefit, emotional hook) instead of features to hook my audience
- Distinct messaging and visual matters.
- Use the P.D.A. framework (Persona, Desire, Awareness) to generate distinct concepts.
- Test out different products in one ad set.
- Considering more about "brand" and giving people a reason to buy products/services that goes beyond just a good price point with solid reviews
- Fewer campaigns/ad sets + more creatives = better AI learning.
On creative volume:
- The was the biggest debate between ordan M and Charley T which went viral on Twitter last week. My take away: assuming PMF, sufficient ad budget and human resource, 'more creative' is always the answer.
- The evidence Jordan presented was fascinating. An ad account that spent $3.5MM in the last 30 days and ran 9,973(!!) ads. That was the scale of creative volume he was talking about. I honestly don’t think regular brands can achieve that but this just shows that volume is one of the most diligent ways to scale.
- However, without putting thoughts into diversity and creative marketing angles, a sheer volume of copy/pasting ads is not going to work either. It screams low quality and feels lazy.
On creative refresh:
- First, ad fatigue is real. Each ad can only cater to so many people. Once the algorithm exhausts the audience, performance drops off. Obviously the better the ad, the bigger the targeted audience.
- Second, for people who don't know which ads will work, the only way to find out is to run more ads. Period.
The rise of AI creatives and how to win with Andromeda.
Zuck launched a bunch of AI tools in Ads Manager that nobody asked for. I’ve heard that over and over again that people find turning off all the AI features actually performs better.
However, AI is an inevitable trend. In the past, marketers and creatives/creators were somewhat divorced. But with the rise of MidJourney, then Nano Bananas, and now Sora, creatives and idea generation are so cheap and so good that the only reason to not have hundreds of creatives is laziness.
A few practical tips about how I use AI:
Who are your customers?
Use ChatGPT's "Deep Research" to automate what used to take weeks, this is to answer the question: Who are your customer personas?
Prompt it to find:
- Target audience (Who)
- Their desires & pain points (What)
- Motivations & objections (Why)
- What language they use (not what you use)
Add this to your playbook → Build your customer personas. Find out at least a few persona that fit into your ICP criteria and describe them like you’ve known them for a long time.
What do your customers want?
This the to answer the question, what's your unique mechanism/process to achieve the desire for your customers?
Create Angles Using:
- What are some USPs of your product?
- What is the desired product outcome?
- How does it help your personas problems?
Concept mapping:
Try desire, painpoint, case studies/testimonials, Objection flipping, Pattern-Interrupts and scroll stopping angles
Creative brief formula:
[Pain/Frustration] + [Big Promise/Relief] + [Speed to delivery or your mechanism/process] + [Drop proof/case study]
What stage are your customer in the buying process?
- Cold audience [TOF Awareness]: They don't know they have a solvable problem. Your ad needs to focus on education.
- Warm Audience [MOF Awareness]: They know something's wrong but not the solution.Your ad needs to introduce a solution.
- Hot Audience [BOF Awareness]: They know they need a solution but not why yours. Your ad needs to show what makes you different or effective.
Put them together & automate the process
Once you describe all these concepts and angles to AI, all you need to do is automate the process. AI can generate so many creative ideas beyond what you can imagine. You just need to learn how to combine the top 3 things and prompt the AI to get results and repeat the process.
e.g. Concept 1: Busy office workers + Need health lifestyle + Warm Audience
e.g. Concept 2: Stay at home moms + Want Energy + Hot Audience
Mix and match these elements, then let AI generate variations. Embrace AI and rebuild your work process.
Lastly I want to call out a few other things:
- Is campaign structure a key influence?
This is controversial. I know a lot of people talk about how they transit multiple ad sets into fewer and more creatives in the ones ad sets, e.g, 1 campaign × 1 ad set × 25 creatives > 5 ad sets × 5 creatives. I just don't think there is a universal conclusion to this. If your creatives and offer and product really work, the influence of campaign structure isn't that big.
- There’s no doubt you are having less control in ads targeting and hence good data infra matters even more.
- For Andromeda to learn well, it needs reliable conversion data.
- High signal integrity (e.g., via Conversions API), good event tagging, and accurate attribution. Weak or noisy data will hurt the AI’s ability to optimize. (Aimerce is here to help!)
- Lastly, really spend some time double-checking your user cookie consent; lots of data loss is coming from that.
Here’s to a smarter, calmer, and more profitable Q4.
- Yiqi