Quick note: your team probably has plenty of creative ideas. What they're missing is a system that remembers what they tested, why, and what it actually told them.

We've seen this pattern play out dozens of times.

A team launches eight new creatives.

Two perform well for a week. Performance dips.

Someone says "we need fresh creatives.”

They launch eight more. Two work. Repeat.

Six months later, they've burned through 80+ assets and can't tell you a single thing they’ve learned.

That looks like testing, but it's closer to gambling with a content budget.

These teams are working hard and producing volume. They just have no system for capturing why something worked, what variable actually moved, or what to iterate on next.

Every round starts from scratch because nothing from the last one carried forward.

And this gets expensive in a specific way: Andromeda rewards creative diversity, but it rewards meaningful diversity.

Throwing ten variations of the same flat-lay with different headlines into an Advantage+ campaign doesn't give Meta's system enough differentiated signal to optimize against. You end up diluting spend across assets that are functionally identical, and Andromeda treats them that way.

The teams we've seen win have figured out how to test with memory, so each round builds on the last one.

(I built a creative testing tracker for exactly this problem. The full template is linked below.)

When we sit down with a team that says "creative testing isn't working,” the issue almost always lives in one of three gaps.

"Let's try UGC" is a direction, not a hypothesis.

"Let's try UGC because our hook rate is below 20% and static ads are getting skipped in under 2 seconds" gives you something you can actually evaluate.

Without a hypothesis, you end up looking at ROAS after the fact and shrugging.

Most teams skip this because it feels like overhead.

But a one-line hypothesis per creative takes 15 seconds and saves you from the most common trap in creative testing: declaring a winner without knowing what it won at.

This is the one that really costs teams.

We've watched people re-test the same angle three times in six months because nobody documented what they'd already tried.

Or, if a creative works, they want to iterate on it, and they can't reconstruct what made it different from the version that flopped.

If your creative history lives in a Slack thread, a Notion doc nobody updates, and someone's memory, you don't have a testing process. You have institutional amnesia.

(The tracker template solves this at the structural level. Every test gets logged with the hypothesis, the variables, the results, and the recommended next move. I'll link it at the bottom.)

Even teams that do track results often stall here.

They know Ad 7 beats Ad 4, but they can't explain why. Was it the hook? The narrative arc? The offer framing? The format itself?

Without that framework, iteration is just guesswork wearing a nicer outfit.

What I've seen work: after each testing round, force a 10-minute read-out where you answer three questions.

That maps directly to the metrics that matter.

You can act on that. You can't act on "the ad didn't perform."

The shift is disciplined, not complicated.

Spaghetti testing looks like this: launch a bunch of creatives, see what sticks, kill the losers, start over.

Structured iteration looks like this: define a hypothesis, test one variable, log the result, use the learning to design the next test, compound your understanding of what your audience responds to over time.

Spaghetti testing is fast but circular.

Structured iteration is slightly slower per round but linear, because each test builds on the last. The gap between the two grows wider every month.

In a post-Andromeda world, this matters more than it used to.

Meta's system evaluates creative relevance dynamically. When you feed it genuinely differentiated assets with different arguments, different emotional angles, different formats, it can find pockets of your audience that respond to each one.

But if your "diverse" creative set is really just five versions of the same message, Andromeda sees through it. Your learning phase drags, delivery concentrates, and cost per result creeps up.

The teams scaling right now are producing intentionally different creative, informed by a documented history of what they've already tried. That documented history is what makes the difference between volume and velocity.

I've put together two things to make this usable for your team.

If your current creative process feels like throwing spaghetti at a wall and hoping something sticks, this replaces the wall with a whiteboard.

Forward this to whoever's managing your creative pipeline.

Get the tracker →

I've seen too many teams struggle with AI implementation because they skip the fundamentals: clear goals and solid training. That's why I wanted to share this AI Training Checklist from You.com.

It helps you build a confident, capable team with practical steps for creating an effective training program, overcoming resistance, and a worksheet to track progress across your org.

AI implementation often goes sideways due to unclear goals and a lack of a clear framework. This AI Training Checklist from You.com pinpoints common pitfalls and guides you to build a capable, confident team that can make the most out of your AI investment.

What you'll get:

Get the checklist.

CBO
Campaign Budget Optimization

Or in this case,
Can’t
Build
On what you don’t track

Because the algorithm optimizes delivery.
Only you can optimize what you learn.

💌 Before you vanish:
Wait a second…

Was this memo great? 🥺🥺
if it landed for you, hit reply and say “Yes”.

More Ad-ventures coming next week!

The Creative Strategist
at The Marketer’s Memo