My exact creative strategy process, start to finish.

Perhaps the most important part of having ads that make you money versus ads that drain your bank account is the creative strategy. Most people agree on that. It's also one of the hardest things to execute well without breaking the bank.

So it's a chicken and egg problem. To make money with ads you have to invest in content. But that's hard to stomach when it feels like the content isn't working, and now you're losing money on the ads and the content. Why is that? What all goes into having a high performing ad creative strategy? What part does execution play?

I remember back in 2018, I was standing in Buenos Aires on the 24th floor of WeWork. I had been traveling and working for a few years, and my creative strategist and I were discussing, "do you think it's possible to actually KNOW what works and doesn't work on Meta Ads?" (At the time we called them Facebook Ads.) We had a program where we created new ads weekly for our clients with their own assets. We were testing messaging, angles. We had a whole "creative matrix" at one point that we tried to execute on. But even doing this, it seemed very random what was actually working and hitting client goals. And it felt like pure luck when something turned out to be scalable.

In fact, the reason we did creative strategy wasn't even because I understood how powerful ad creatives were at the time. It was more because, like many agencies or ad managers, we thought that "testing" was the most important part. The more tests you do, the higher the chance you start hitting your performance goals. The problem is, when those tests aren't focused around detailed learnings, and when they don't move in the direction of what's working consistently, they actually waste ad budget.

I say with confidence now, with the right creative strategy that can move quickly enough in the direction of what's working, you can indeed learn what works for Meta Ads and build upon it. Scale should come from building upon what's working while at the same time diversifying the looks and feels around the core messaging that's working. Not just "more ads," "more tests."

So in this post, I'm going to share start to finish my exact creative strategy process. I'll try to share it in the most repeatable way possible so you can copy and paste.

Step 1: Performance Report

The most important part of your performance report is NOT the ROAS or CPA/CAC or CPL. It is spend. I learned this the hard way. When you fight what the algorithm spends on, you assume it's getting things wrong. It rarely is.

So let's talk about how the algorithm spends on new creatives for a minute. You probably think of your "audience" as an Ad Set. That's where we set the audience after all. But the real audience is built per ad by the algorithm. It will observe who is interacting with that ad and then create a small subset of your total addressable audience to show that particular ad to. As it learns more and more, the data behind that ad builds and that audience gets tighter and tighter.

Data isn't just about what's working. It's about what isn't working. The algorithm prefers when it confidently knows who will respond positively to your ad, because it can spend your budget most efficiently that way. With a new ad, it doesn't know who to show the ad to. Even within your total audience, most people won't react or buy. And it's expensive to show your ads to everyone. That's what happens when you force spend to a new ad where the algorithm doesn't have enough data to create the subset of the audience. Occasionally it does work. But most of the time it just spikes the costs behind that ad.

If you suspect the algorithm is "getting it wrong," you can try forcing spend to an ad creative in a new campaign. We call this "warming up your ads." Really it's just forcing the algorithm to spend before it's confident about who to show the ad to. To be honest, prior to Advantage+ and Andromeda, I was never a fan of this. Because you are testing in one environment (warmup campaign), then trying to perform in another environment (evergreen campaign). The data in your test campaign might be different than the data in your evergreen campaign, and sometimes one doesn't translate to the other.

With Andromeda, in the beginning it was very difficult to test new ads within the evergreen ecosystem, but that has now subsided some. And I want to encourage you to detach from your idea of what "testing ads" really means. When we "test ads" it's no longer about finding a winner. It's about keeping an ecosystem of ads within one of our campaigns fresh and high performing. When you run a mature ad program, there should always be new ads going in and ads not spending well coming out of your campaigns. Ideally on a weekly basis (sometimes more frequently for high spend ad accounts).

So back to the performance report. If you are introducing new ad creatives weekly to your campaigns, you will see over time (usually 30 to 90 days depending on your ad account spend levels and your AOV) that new ads do start picking up spend.

Step 2: Hypothesize

There will always be a "why" attached to your ads that pick up spend versus your ads that don't. It could be the hook, it could be a particular creator, the editing style, the messaging, the script, the energy, the headline. It could be anything. But there is a reason.

Don't try to fight what spends. Learn from it. If you're confused by what the algorithm is spending on, that confusion is the signal. There's a why sitting underneath it that you don't understand yet. So the first thing you want to observe is what are your top 5 spending ads. Make sure to separate your top of funnel from bottom of funnel ads.

If you are running a large program, you might have ads getting a lot of spend with higher hook rates, and you might have ads getting a lot of spend with lower CPL, CAC, or higher ROAS. This is an added layer that you can start observing. Some ads will play the part of introducing the brand and some will play the part of bringing home the lead or sale. These two roles that your ads play should be understood differently. BOTH are valuable. Don't discount the ads that are grabbing attention and spending well just because some of them might not show the best end results. Again, this is mostly relevant for large accounts ($100K+/mo spend). The more you can think about your ads as an ecosystem versus one-off ads, the more successful you'll be working with the modern algorithms. This is not only true on Meta, it's also true on other social platforms like TikTok and even somewhat on YouTube now (though less so).

My team and I have long, deliberately messy meetings discussing the "why" behind ads. It's a creative process, and the more you allow yourself to think through possibilities, digging into each of those top 5 ads second by second, the closer you tend to get toward the "why." Typically I have my media buyer, myself (chief strategist), and my video editor on these calls. This ties together different viewpoints on what makes up the ad.

Step 3: Iterate

The hypotheses that come out of the above meeting are executed on with either new assets and/or new edits. Sometimes we need additional assets and we put that in motion with our input calendar. But sometimes it's as simple as trying a hook style on a few other videos we tested that didn't do as well. Does that improve the performance? Sometimes it's about caption style, so we add a caption style to some videos that didn't do as well. And occasionally it's something more like "we need more outdoor footage," or "that creator seems to do well consistently, let's try this other script with them."

"How can we get a new version of this ready to test by next week?" When you put a time constraint on it, the thinking changes.

We try to move as close to weekly as possible. "How can we get more footage like that in the next week?" That should be the question. It stops being about big steps forward and becomes about small ones. This is exactly how you learn, because it creates an environment of micro-testing but in a simple way, not a complicated way.

Step 4: Diversify

Within the same meeting where we hypothesize, we leave some time to diversify. With Andromeda this is more important than ever. NEW concepts, not just iterating on concepts that work. My favorite way to get ideas is just keeping my eyes open as I go about my day. We consume SO MUCH media in our days. Whether it's ads on TV, ads out and about, signs in the neighborhood, the feeds with ads and organic posts, there are ideas everywhere.

It's important not to limit yourself to getting ideas from your own industry. One of the best ways to truly be innovative in your industry is to take from other industries. Or related industries. For example, if you are a fitness brand, pull inspiration from fitness apparel brands, gyms, fitness influencers, famous athlete brands, professional sports teams, and supplement brands. It's not just about copying a video or static image. Try to get inspiration in elements (editing styles, caption styles, story types, special effects, image design, messaging angles), similar to how we build the creative library in raw elements versus in whole videos or statics.

Having an ecosystem of diversified assets is what feeds Andromeda. So unlike many brands right now, if you follow this system you won't have an issue, and in fact you'll utilize the power of Andromeda well, which will improve your overall account health and performance significantly.

Step 5: Assets

I have this step as number 5 because I want you to be building your asset library based on data and defined inspiration, not randomly. But the truth is, this should be happening simultaneously to all the above, consistently. It should be disconnected from the other steps and repeating consistently.

Here's the thing: most brands are gathering assets all wrong. They are trying to get assets as a one-stop-shop. The closer they can get to the "finished video" the better, because they feel this saves them time. But what it really does is make your content gathering process slow, inefficient, and EXPENSIVE.

If your creative agency or your creators are sending you finished products, you can no longer use the assets for other projects (if they are done well). Because the clips will be 1 to 2 seconds long versus the raw, which was probably 10 to 15 seconds at least. They'll have captions overlaid, so suddenly you can't use that clip as b-roll. It can only go with that script and that audio. That one decision makes every asset roughly 20X more expensive. It's also why, even though you're paying for a mountain of content, your ad team is always asking for more.

When you build your asset library RAW instead of edited, you are able to reuse clips 100 times over. You can use them as b-roll, with or without audio, different music overlays, different caption styles, different hook styles, different editing styles, transitions, overlays. You can make raw footage look completely different with all of the above. You lose all of this flexibility when you get a finished video. Another issue: even if you can splice out the clips, the quality of the edited exported version is sometimes lower than the original raw footage. For example, if you want to pan in or out, or use it beside another clip that is higher quality, immediately it's more noticeable.

So separate out how you are gathering your content. It's going to be cheaper on the front end and easier to gather. Creators are happy to send raw footage, and the turnaround time is usually way faster. If you pay for a professional video shoot, it reduces the time frame significantly on getting the content back ready to use. You want creators who are comfortable shooting content. You want a professional videographer whose skill is getting the high quality shots with the right lighting. But you want an editor, or a combo of media buyer / creative director and editor, who know how to SELL by turning the content into an ad. So remember two things from this section:

1. Build your asset library with RAW, not edited videos.
2. Disconnect the asset library building from the rest of the process. It will go much faster that way.

Step 6: Design

Let's talk about design first. AI has made this process incredibly easy. And it's a great way to have diversity of looks and feels within your ad account without a ton of effort. There are still a few important things to consider with designing static image ads.

The most important part is the messaging. And we love using static image ads to test messaging angles more than videos. The reason is to control variables. With video, there are other things happening while the user is reading the text, so it's harder to know for sure if that messaging was the driver of performance. But on a static image ad, the confidence in the results of messaging goes up, assuming imagery is a constant.

Remember that just because you have text on your ad, it doesn't mean people will read it. And just because you have that text in a certain order, it doesn't mean they'll read it in that order. The eye bounces around based on typography, so how big or "in your face" the text is. You'll usually have imagery too, so ask yourself: is the user noticing the image first and then the text, or the other way around? The reason this is important is so you can understand what story they are gathering from the ad. You might be trying to tell one story, but if they mix up the parts too much, sometimes they will get a different story.

Even though we can't measure it, your static image ads still have a hook, and they still have a hook rate. We just don't know what it is. That's why it's very important to think through what you think the hook is. Is it the image? Is it the text? What part about the ad will catch the user's first glance? Is it potent enough to stop the scroll? It's interesting, because Meta knows the hook rate of static images. So also observe what gets the most spend. That is usually aligned on the video side with higher hook rates, so you can assume it will be similar on static images. What spends is performing best, and hook rate is a huge part of that. If you don't hook the user, you don't have a chance to perform well in the first place.

Step 7: Video Editing

This is my favorite part of the entire process. It's where you can create almost anything if you have a substantial raw asset library. I could spend all day talking about editing videos for ads. But for the purposes of this post, I'll keep it as succinct as possible.

Performance is not about a video. It's about all the parts of the video in 2 to 3 second chunks. Because you are getting attention, then keeping attention, then keeping it again, then keeping it again. Attention is not guaranteed at any part of your video. It does get easier to keep it after the first 6 to 7 seconds. I try to think about my ads like this:

You have so many elements to layer into each one of these phases. Editing effects, transitions, transition styles, text overlay, text size, color, weight, font, music style, music aligned with editing transitions or effects, audio, captions, caption styles, AND yes, also the visuals of the actual clips. This is why it's SO important to build your asset library raw, so you can USE your content with all of these tools we have on the editing side.

I separate my editors into two categories. First, the high quality (and, yes, expensive) editors. They're the ones creating new concepts, helping on the ideation. And then the volume editors, who can take high performing ads and version them, but aren't as good at creating new concepts. They are much more affordable though, so this cuts down significantly on editing costs overall.

I give both sets of editors volume goals per week. Their performance goals are built on executing a certain number of ads per week, or sometimes we set the goals biweekly or monthly to be delivered weekly. This separates them from the rest of the assembly line and allows them to focus on what they do best. In the event there isn't a defined brief, they have access to the assets in the raw asset library and they have access to the performance reports, so we have them operate independently when needed. This keeps the system moving even when the media team hasn't gotten perfectly conclusive data yet, or maybe it was a holiday which threw off the stats. That is best case scenario, where you can give the team a little breather on fully defined briefs while keeping the machine moving and the content fresh. They get to create without pressure.

Step 8: Approvals

I work hard with my team to remove approval processes with specific SOPs. The more responsible and autonomous each team member is, the smoother and faster the process goes. When editors or media buyers think someone else is approving, they aren't as likely to take responsibility themselves. So I like giving everyone on the team full responsibility for typos, anything that renders wrong, and so on. Make it everyone's problem and responsibility.

Step 9: Media Team / Testing

Just like many of the other steps above, I like separating this step from the rest. The media team is responsible to launch new ads in every Ad Set every week on every account. No matter what. To make sure this is possible, we try to overshoot on the editing goals so the media team always has more than they need. In the event a week goes by without new videos or images for whatever reason, we have extra ready to go.

Another thing we do is pull from organic content. Now that Meta is serving organic Reels based on purchase data, these two areas of Meta are more connected than ever. If an organic Reel did well, chances are it can do well as an ad. So this is also at my media team's disposal, within the constructs of client approvals.

When the team launches a new ad or group of ads, they are also optimizing out ads that didn't perform. It's important to know how long it takes new ads to pick up spend or not in your ad account. If they pick up spend within 3 to 4 days, you'll be ready to pause out non-performers weekly. But sometimes it takes a few days for them to bake prior to picking up spend. In this scenario, on video ads, the hook rate is normally the indicator. If an ad has a very low hook rate the first week, it's almost never going to pick up spend. So my team learns these nuances in each account, defines them, and optimizes out low performers as they launch new options for the algorithm to spend on.

And that brings us back to Step 1. Rinse and repeat. So there are 9 steps to creating your creative strategy and execution program that feeds Meta Ads with high quality, high volume content, without breaking the bank: build a RAW asset library, run two layers of video editors, and use AI for static images.

Frequently asked questions