If I Took Over Your Creative Strategy Tomorrow, Here's How I'd Structure Your Tests

Last month I came across an account running 40K in monthly budget with fifteen campaigns cannibalizing each other. Creatives launched more or less at random, budget scattered everywhere, and the founder asking me where to even start. So I'm going to answer you like you're asking me the question today. If I took over your creative strategy tomorrow morning, here's exactly what I'd do.

No theoretical fluff, just the structure I apply to the accounts we manage, step by step.

First thing, I consolidate your account

Before we talk creative, I look at your structure. And 8 times out of 10, it's a mess. Fifteen campaigns, ad sets cannibalizing each other, budget scattered everywhere. You learn nothing in those conditions because the algo doesn't have enough signal on each line.

So I tear it all down and start simple. One single campaign, four ad sets. One per stage of the buying journey.

The first ad set targets people discovering you for the first time, those who don't know you, who didn't ask for anything. The second catches those starting to show interest, who click, who watch. By the third, they're comparing, hesitating, adding to cart and closing the tab. The last one is the purchase.

Four moments, and at each one people expect completely different things. You don't talk the same way to someone just discovering you as to someone who's had your product sitting in their cart for three days. And that's exactly the mistake I see everywhere, a single creative blasted at everyone hoping it sticks.

If your budget is tight, start with just one ad set, the top-of-funnel trigger. It's the ad set that brings you the most fresh volume, so start there. You don't need fifteen campaigns when you're running low volume. You want a lean setup where every euro brings back usable data, not some overengineered machine where you no longer know what's working.

3 creatives per test, no more

Once the structure is in place, I produce 3 creatives per test, spread across three formats: a video (UGC or generated), a static, a carousel. One single concept behind all three. I do it this way because format changes everything. The same message that flops as a static can crush as UGC. We see it every single time.

If you've got a bit more budget and want to run 5 creatives in the test, I shift the mix. Two videos, two statics, a GIF or a carousel. But don't start trying to test 12 things at once. You'll dilute your budget, no creative will get enough volume for us to call it, and you'll be no further ahead.

The principle is to test the three formats together with the same budget on each, to keep the reading clean. Say you end up with the UGC performing way better than the other two. Great. Then you make 3 more UGC pieces with 3 different angles, you test, and you find the concept that really works. It's iterative, each round you refine.

This reflex of wanting to produce too much at once is a trap. I talk about it in this article on how many ads to run at launch. Fewer at the start, more creativity after.

The micro-moment, writing for a situation, not a persona

Now the thing I've never seen anyone around me do. When I write the top-of-funnel creatives, I don't stress about avatars, personas, who we're talking to. It's useless at this stage.

I pick a single micro-moment I want to cover, and I make all my creatives speak to that exact moment.

An example. Picture a snack brand. The micro-moment could be "I hate settling in for a movie and realizing there's nothing good to snack on." That's not an age bracket or a zip code. It's a lived situation, a daily friction people recognize instantly.

From that one moment, I build out my three formats.

A UGC video of someone griping in front of their empty cupboard. A static with a hook like "the day I decided to throw out all my crappy snacks." A carousel, "the 7 worst snacks I ate on my couch this week, so you don't have to."

One single micro-moment, but declined across three formats, each attacking from a different angle. And the hook matters enormously here. The first three seconds of a video, the first line of a static, that's what makes someone stop or scroll. You can have the best product in the world, if your hook doesn't hit the right moment, nobody sees you.

Where the concepts come from

The real question comes before the writing, where do I go to find the idea. And there, it's always client data that decides.

I look at what's working right now on competitive monitoring tools like Trendtrack. I dig through Reddit to see what people are talking about, recent threads, the frustrations that keep coming up. And above all, I read customer reviews. The real ones. Often they list every other product they tried before yours, and that hands you golden angles.

Every month on the accounts we manage, we benchmark the concepts that are working, in the niche and among competitors. That gives us a constant stream of new shoot ideas to test. Because the brands that scale best are the ones that produce the most content. Always.

Scale what works horizontally

Once the test is live, I let it run, I collect clean data, and I call it. What doesn't connect, I kill. What works, I scale. But horizontally, not by dumping budget blindly.

Scaling horizontally means I take the winning concept and decline it again.

Back to the "7 worst snacks of the week" carousel that crushed. Instead of relaunching the same carousel with more budget, I keep the 7-snacks concept and run it elsewhere. A narrative video about the gang of 7 bad snacks. Or a provocative static, "nothing's worse than these 7, prove us wrong." And why not a GIF done like a mini-game.

Same core idea, but each format tells it differently. That's how you squeeze everything out of a winning concept instead of burning it in three days.

On the ad sets, I keep it simple. One testing ad set, one scaling ad set. I put all new batches in testing. As soon as a creative overspends and performs, I move it into scaling. That way you always know where each creative stands.

And to judge, I don't look at just one metric. CTR first, because it tells you whether your hook catches the eye. But I don't stop at the click. I look at what happens after, add-to-cart rate, conversion rate. A creative that pulls tons of clicks but no sales is a trap, it burns budget and tricks you into thinking it works. That whole post-click part, your site taking over, is just as important as the creative itself. A bad site kills the energy your ads just created.

One last thing on variety

Something that comes up all the time. People duplicate their winning creatives, they relaunch the exact same ones hoping they'll hold. It doesn't work. The algo loves variety, it needs new concepts to keep delivering well. Duplicating creatives is a false good idea, you stall instead of progressing.

At its core, creative is a game of combinations. When your structure is clean up top, you've got free rein down below to test anything you want on concepts and formats. Order matters. If the structure is a mess, you can produce the best creatives in the world and you'll never know which ones work.