How $10M+ ARR apps use AI in creative production

AI is changing how app teams research, produce, and test ad creatives. But creating more ads is only useful if you improve your chances of finding winners.

At Spiral, we work with some of the best app teams and see firsthand how they use AI in paid growth. The biggest challenge keeps coming up: teams are producing more creatives, but most still don’t perform.

So what are the teams getting it right doing differently?

This post covers five practical ways they use AI, from analyzing creative patterns and testing new markets to building automated workflows with live performance data. I originally shared these takeaways at the Snapchat Performance Summit in Warsaw. Here’s what they mean for your app’s growth.

‍

What has changed in paid growth

First, algorithms have become smarter, and your creative now does much of the targeting. Different hooks, problems, and messages help you reach different audiences. That makes creative the biggest driver of app growth.

Second, campaign management is increasingly automated. Running ads on Meta TikTok or Applovin requires less manual setup and optimization than it used to. You still need to understand your numbers, but the platforms handle more of the execution.

Third, ad libraries have made competitor research much more transparent. This year, Meta and TikTok introduced sorting by impressions. This means you now know top performing ads of any of your competitors and can explore this data in dynamic. Once an idea gets traction, expect others to copy it.

Finally, the app market is a bloody red ocean. In almost any category, you’re competing with dozens, hundreds, or even thousands of apps solving similar problems for the same audience.

So where apps are struggling with paid growth?

During onboarding at Spiral, we ask customers about their main creative bottleneck. Here’s what our data shows:

• 48% said most new ads don’t perform.
• 18% said creative production is too slow.
• 14% weren’t sure what to test next.
• 12% struggled to understand what drives performance.
• 4% said winning ads burn out too quickly.

The biggest bottleneck by far is we make new ads, but most of them don’t work.

So why do new ads not perform?

Power law distribution.

Creative performance follows a power law distribution. Around 3% of ads drive 80% of revenue for most mobile apps on Meta

So the most important question in paid growth is: how can you improve win rate with AI?

The five approaches I’ll share next aren’t just theory. They’re how top app teams are already scaling their creative production with AI

Five ways top app growth teams use AI in creatives

1. Centralize your product context

This is the foundation. Without it, everything else is useless. Put your audience, product benefits, customer feedback, previous tests, and successful ads in one shared place. Keep it updated

“Write an ad for a fitness app” gives AI no context to work with. Explain who uses it, why they chose it, what they struggle with, and which promises your product can deliver

I recommend turning this into a Claude or Codex skill, so your agents automatically use it whenever you create a new ad, saving you a ton of time

2. Analyze creatives with AI tags

Use AI to break down ads by hook, concept, audience, emotion, and more. Find the patterns behind your winning creatives. I recommend Gemini models for videos, but you can also try Jev for static ads

Previously, many teams relied on ad naming templates and manual tags. Now, AI can give you much richer information about your ads in a fraction of the time, helping you uncover patterns behind them

Once you have your AI tags, you can analyze which tags are associated with your best-performing ads and which combinations drive the strongest results. Use these insights to guide your next creative decisions

3. Localize creatives to smaller markets

Adapt creatives to local languages and cultures in bulk, including smaller, less competitive markets. People in the US know about calorie tracking. People in Bulgaria might not

Before AI, many teams wouldn’t even consider localizing ads for certain languages. Now, I’m seeing more teams launch in smaller markets because AI has become so good at translation

AI also changes how teams approach ad translation. Previously, you needed a clear winning ad before investing in translations. Now, you can test multiple languages from the start and discover which angles perform best in each local market

4. Set up automated actions

This is more advanced, but is crucial for real scaling without increasing your headcount. Trigger AI workflows based on creative performance data from your ad accounts

Three automations that I see already work:

1. Translate winners. Trigger translations when a new winner emerges. Translate the voiceover and captions, and adapt the messaging and visuals to the local context.
2. Generate new hooks. Trigger new hook variations for promising ads with a good conversion rate but a low hook rate
3. Swap AI avatars. Trigger avatar variations when an AI UGC ad performs well. Combine this with translations when entering a new market to test which presenters resonate with the local audience

5. Give agents access to live creative data

And finally, the one I’m personally most excited about, and believe is the most powerful = MCPs. Power AI workflows with the enriched creative intelligence data and trends.

Essentially, instead of attaching files and manually searching ad libraries, your agent can access creative data with trends, winning ads, and transcripts created for you.

Examples: "Find trending Education ads and break down their strongest hooks" or "Notify me when a competitor finds a winning concept and give ideas how to adapt it for my app".

There are ton of nuances to get it right. Try Spiral MCP to get app market intelligence, creative generation, and performance data into your AI agents. We track 7,000+ apps actively advertising on Meta and TikTok

That's it. Remember that paid growth is a never ending proccess. A winning ad lives 2 weeks on average, and usually you need to test hundreds of ads to find your top 3% winner

Let the ROAS be with you!

‍