In the realm of advertising, numerous market players have embraced the power of AI to meet their challenges. AdChampagne‘s CEO Yaroslav Pankov and Head of Media Buying Artem Bezmenov explain how AI fuels agency development and drives customer business growth.

The art of generating creatives: Our solution

On the stage of performance, a significant portion of our efforts centers on hypothesis testing, optimizing purchases, and continuously updating creatives. With the average lifespan of a typical creative lasting just a week, certain customer categories require the generation of dozens of video ads per day. In order to avoid burning out from constantly coming up with new ideas, we came up with an innovative solution — «‎NeuroCreo», ‎harnessing the power of AI in the creative process.

The platform is based on the MidJourney model, which was the easiest to integrate, as well as the most relevant and effective AI solution. We trained MidJourney on a diverse range of ready-made creatives from various categories. Furthermore, our team have systematized “prompts” that allow us to create suitable ads much faster and require less involvement.

Prompt — any form of input or query that a user gives to AI in order to get a specific response

«‎NeuroCreo» promptly considers the format and technical requirements for each creative. So, media buyers simply need to specify the keywords and receive a ready-to-launch ad. Currently, AI-generated creatives are showing outstanding results in the clothing category. Our gaming team use this technology for generating characters, backgrounds, locations, and scriptwriting.

Click on image for full size

Source: AdChampagne

Currently, the majority of images from «‎NeuroCreo» undergo revision by our designers. And we have already reduced the time for the creative process by 80%. Previously, a project would take a whole week to complete, but now we accomplish it in just two days. We have also implemented an in-house text generation solution, which has greatly reduced the workload of our copywriters.

Campaign analytics: Model-building process

During our collaborations with a wide range of clients, we have accumulated a great amount of data. One of our main goals is to leverage it more effectively. Our team successfully completed the initial phases of data collection and now delving into the model-building process. The key objectives of our platform revolve around achieving several important outcomes.

Firstly, we aim to optimize bids, ensuring that our clients can make the most out of their advertising budgets. Secondly, we strive to enhance conversions, maximizing the return on their marketing efforts. Lastly, we seek to organize and systematize traffic sources based on their cost-effectiveness in attracting valuable leads.

Extra monetization: Personal assistant bot

Thanks to the resources within our team, we were able to expand beyond just advertising services and venture into creating an alternative product. We have launched the MVP app Yollu – a service offering personalized assistant bots in various professional fields. The product simplifies access to AI for educators, accountants, lawyers, builders, linguists, cooks, IT specialists, psychologists, and even cultural critics.

Yollu app

Click on image for full size

Source: AdChampagne

Users can easily submit text-based requests and receive prompt responses from bots trained by different neural networks, each focusing on specific categories. These responses can be saved to bookmarks and then utilized through pre-made templates. In the free trial, the number of requests is limited, but it’s possible to remove the limit by subscribing.

Growth points and prospects: The future of AI

Presently, we are testing a solution that analyzes data of completed campaigns, and based on them allows you to automatically launch new ones and predict the necessary budget for promotion. The tests demonstrated a 2X reduction in CPA.

We partially use it in development for elementary tasks — the AI code-writing approach relies on simplistic methods that overlook context and architecture, rendering it unsuitable for cross-platform development. The speed of implementation of simple architectural solutions has already increased by 3 times.

We would also like to integrate AI for video creation, but so far there is no technology on the market for producing high-quality video and audio products. As of now, manually crafted videos continue to outperform AI-generated ones in our campaigns.

Conclusion

Thus, with the help of AI, certain tasks can be solved much faster. However, it’s essential to know that a significant amount of time is spent on training the teams and neural networks. AI often provides initial ideas, but don’t forget they need further improvement. That’s why our team follow a practical approach: we use AI only when it’s genuinely needed. We apply this technology when it reduces costs for the clients and optimizes our resources. And, of course, continuously experiment and closely monitor AI’s advancements worldwide.

At the moment, AI is doing up to 5,000 requests daily for different agency tasks. If we summarize our experience, it’s been a valuable asset in many ways:

  • Cuts down the time and expenses needed for producing creatives, by about 80%
  • Allows to generate more creatives, reaching a broader audience and more traffic sources
  • Unloads from routine tasks, freeing up time for learning and creativity
  • Brings new opportunities for monetizing and growing our business.
  • Enhances our team skills and gains a competitive edge in the market