The short-form video market is booming. With a 10% CAGR, it is projected to reach $5.63B by 2036. I spotted this trend three years ago in Asian countries, and along with a co-founder, Bogdan Nesvit, we launched My Drama, a production of short engaging series.
Our company has reached $90M ARR, catching the attention of other startup founders and investors. But if you think it's all about a promising niche and a quality product, don't rush into this business just to make easy money. Startups like Quibi raise enormous investments and close down after a year. Why? Because a huge part of success in this industry depends on the distribution system. This is a crucial thing we've built perfectly that allows us to grow rapidly.
In five years of developing products within the media tech company HOLYWATER, we've built our own AI marketing machine. Having spent over a million dollars on it, we now have a data-driven assistant for generating, implementing, and scaling creative concepts.
Want to know the real story about how to promote short series to billions of people? Here's how we do it at My Drama.
In-house AI engine for generating concepts
Every month, My Drama generates more than 15,000 original concepts that are localized into 12-15 languages, totaling over 200,000 creatives launched. No, we don’t have hundreds of marketers for this. Our marketing machine doesn’t operate only at scale — it operates smart. We combine human resources with internally created AI tools.
For example, we've developed a solution that identifies viral elements from the series, cuts them out, combines them, and adds titles. As a result, we receive a finished concept.
How does all that automation look in practice? Take a look at the marketing team structure below
While the production team brings creative marketing ideas to life, AI automation specialists ensure that most concepts are generated using custom-made solutions.
Everything is built on data. Each creative is analysed, revealing which hooks work. The system learns from this information and improves, so each new wave of creatives performs better than the last.
What started as a million-dollar investment has become our growth engine's cornerstone.
Combining AI tools with human creativity
We continue beyond just using internal AI to generate creative concepts. Many AI solutions on the market already automate video content processing, and we are leveraging them as well. Some of them include:
- Stable Diffusion and Midjourney for visual asset generation
- RunwayML for advanced video manipulation
- Leonardo AI for artistic elements
- LAMA for smart content adaptation
- Concepter for ideation assistance.
But technology alone isn't enough — our secret lies in the collaboration between AI and human creativity. Our creative team handles complex conceptual work that requires nuanced understanding of storytelling and emotional resonance — things AI still can't fully replicate.
Data-Driven decision making
Each creative concept is backed by hard data. We analyse all interactions to improve the AI machine and the performance of our creatives.
Among the key metrics we look at:
- Early engagement rate –– whether the creative holds on for the first 5 seconds and converts to watching the video to the end. If not, we analyze the drop-off point.
- Performance metrics: cost per completed view, return on ad spend (ROAS), audience retention, cross-platform performance comparison.
To improve these metrics, we constantly research our audience. Our goal is to understand what users want to achieve by watching short series. For some, it's an escape from reality; for others, it's more for chilling out. As soon as we determine the audience's pain, we consider ways to broaden their reach.
Automated media buying
To test over 100,000 creatives per month, we use in-house solutions to launch campaigns on social platforms like Meta, TikTok, Integral, Liftoff, and others. Yes, everything is again built on strong analytics, forecasts, and AI tools.
We spend $6M monthly on advertising running different formats like video, banners, and playable ads. With advanced prediction models, we are sure that our payback is within our expectations.
The approach for each channel differs according to audience preferences and platform rules. For example, for TikTok, we create short videos on viral topics. Our team monitors trends and adds them to advertising creatives. Additionally, we run thousands of tests of different marketing materials: landing pages, headlines, and app store pages.
Here you can take a look at our media distribution cycle.
The gamers' audience is a different matter. Ads during the game are watched until the end because they get a bonus for it. They enjoy these things! Here, we can move away from the concept of the first hook and fulfill the story better as we have more time.
Our next milestone is to scale our creative output 2-3 times and produce more viral content. We already have cases of viral videos that reached millions of views with zero promotional spending. Call it bold, but we want our creatives to become trends and generate 10+ million organic views and engagement! I’m sure it's possible soon.
Anatolii Kasianov is the co-founder & CTO atHOLYWATER.
