Ads 2.0

Meta just rolled out Andromeda - and it changes everything about how ads work.

For years, the algorithm has been a ‘king-of-the-hill’ contest. You’d give it a few ads, it would quickly find the strongest one, and then pour your entire budget into that single winner.

But with the sheer volume of ads being uploaded daily, the old model broke.

Andromeda is ads 2.0.

It’s a completely different retrieval algorithm, built on far more powerful infrastructure.

Instead of finding one ad to show everyone, it acts like a matchmaker. It takes your entire portfolio of different ads and seeks out specific pockets of the audience that will resonate with each unique message.

This is why a slightly different hook isn't enough. And the old way of doing lots of small iterations will no longer work. The system is now actively looking for fundamentally different concepts to match with different people.

This may seem like a problem, but if you’re a good marketer, you’ll understand that it’s a huge opportunity, because most people will not adapt.

The main ways that you have to adapt are:

1. Prioritise creative diversity. Less variations and more concepts. Before, 80/20 rule was 80% variations, 20% new concepts. The opposite is true now.

2. Understand that the creative is the targeting. With image, video and text AI models, meta can understand everything in your ad, and then work out the audience that’s best to show the ad to.

3. Quality over quantity. Before, it was important to eek out 1-2% improvements in ROAS by testing lots of iterations. Now, it’s important to really focus on quality, and make sure that you’re making unique concepts.

So, the question becomes: how do you create these truly 'different' concepts systematically? I've tried to break it into a simple 3-variable framework.

Format x Desire x Awareness Level

Format is how you deliver the message:

• VSL, static image, short-form podcast ad, UGC, carousel, testimonial etc…

Desire is what the person wants. People buy transformations, not products. You need to hit their core motivation.

Example (for a Fitness Coach):

• Desire X (Health): "I want to have more energy to play with my kids and stop feeling tired all the time."

• Desire Y (Relationships/Status): "I want to feel confident and attractive in my clothes again for date night."

• Desire Z (Health/Performance): "I want to finally run that 5k without stopping."

The ad focused on "energy for your kids" will have a totally different emotional pull than the one focused on "feeling confident for date night."

Awareness Level

This is the "Where." Where is your persona on their customer journey?

What do they already know about the problem, the solutions, and you? (Props to Eugene Schwartz for this concept).

Example:

1. Unaware: They don't even realize their constant tiredness is a problem they can fix. Your ad needs to educate them. (e.g., "Feeling sluggish after lunch every day? It might not be your fault.")

2. Problem Aware: They know they're out of shape, but don't know what the solution is. Your ad needs to introduce a solution. (e.g., "Tired of gym routines that don't work? There's a better way for busy professionals.")

3. Solution Aware: They know they need a workout program, but they don't know why yours is the best. Your ad needs to differentiate. (e.g., "Here's why our 30-minute workout is more effective than 2 hours at the gym.")

Now, you just combine the variables to create your unique ad concepts.

You're no longer just tweaking a single message. You're creating a portfolio of ads that speak to different people with different goals at different stages of their journey. This is what Meta's AI wants to see.

Action Steps

1. Start with 4-6 different creative concepts in one campaign, and in one ad set.

   1. The goal of your initial 4-6 ads is to find the winning themes. You are making your best-educated guesses across a range of angles and awareness stages to see what gets the best signal from the market.

   2. You want to consolidate everything to give the algorithm the maximum amount of data and flexibility. No ad set targeting needed. Even with wildly different personas, you would still use one ad set, with targeting completely open.

   3. 1 ad = one ad concept = 1 creative (either video or static)

2. Set a sufficient budget to test your creatives. Daily budget should be 3x CPA and should be enough to properly test your different creatives.

3. First 5-7 days: Don't touch anything. Let the campaign exit the learning phase and give Meta enough time to test all the creatives.

4. After a week, look at the results. You'll likely see 1-3 concepts that are getting the best results.

5. Now, you'll do two things. First, iterate on your winners (e.g., if the 'Busy Pro' persona is working, create a new ad for them with a different hook or image). Second, introduce 2-3 brand new "wildcard" concepts from your F.D.A. framework to keep testing fresh ideas.

6. At the same time, you can turn off the 2-3 ads that are clear losers (high spend with no results, terrible CTR, etc.).