How I Use AI in Media Buy

AI has transformed many industries, including Affiliate Marketing. Even at the upcoming Affiliate World Asia I’m attending, the app’s schedule features dedicated AI tracks, and YouTube is filled with discussions about how AI is changing Media Buy.

Before AI emerged, when people asked me about this industry, I always emphasized the importance of English and programming skills. While they’re still important, they’re no longer absolutely essential. Of course, if your English is decent and you know a bit of programming, AI will only make you more powerful.

If you’re looking to enter this industry, or you’re already in it but haven’t been using AI much, you might be missing out on tools that could double your productivity. This article will introduce how I apply AI in my daily Media Buy work and the benefits it brings. I hope it inspires you.

Translation

In the past, when running campaigns in multiple languages like German, Italian, French, or Spanish, I faced a challenge. English isn’t even my native language, and I could clearly detect the awkwardness when using translation software, let alone how stiff translations into those languages would be. But after AI emerged, when I throw the same text to AI for translation, the quality improves dramatically. The sentences flow naturally and are no longer rigid literal translations.

If I have an offer performing well in the English-speaking world and want to expand it to test in other languages, I just need to send the landing page code or copy to AI and ask it to translate into a specific language. It can do it in seconds, which was unimaginable before. This gives me tremendous expansion opportunities. I’m no longer limited to just the English market - as long as there’s an offer and I want to do it, I can.

Ad Copywriting

The same principle applies to ad copy. In the past, when I wrote copy, headlines, and descriptions, I mainly copied and modified existing content. I had no choice because English isn’t my native language and I was afraid of making mistakes. Now it’s different. I can directly tell AI my ideas, have it write for me, then I select and fine-tune to meet my final needs. I can even extract what I consider excellent copy, then have AI analyze and improve it. Through continuous dialogue with AI, I polish my ad copy.

Going further, I tell AI the results after running these ad copies for a while and have it analyze why certain copies performed exceptionally well and why others fell short of expectations. Then I iterate and create subsequent ad copy. I think this step is important because AI models now have global memory, which means I’m growing together with my AI assistant. As it gets better, I get better too.

Landing Page Content Creation

Previously, I would find landing pages through spy tools, scrape them, and modify them. I basically didn’t touch the content because I didn’t know how to make it better. Now it’s different. After finding good content, I throw it to AI and have it analyze the landing page’s structure and content, its strengths and weaknesses. Then I express my views, evaluate the improved structure together with AI, and finally apply this structure to my new offer content. Everything looks much smoother.

The landing page becomes unique, containing my ideas and content, no longer cookie-cutter.

People always say don’t be a copycat, but without being a copycat, I found it hard to move forward. Now with AI, I can more easily avoid being a copycat, and Media Buy has become less boring.

Code Development and Tooling/Automation

Although I’m a programmer, I mainly do iOS software development, not web frontend or backend. So in practical Media Buy applications, it doesn’t help much, but the advantage is I can read code and have some programming mindset, even with different programming languages. That’s why I say it’s better if you know a little bit of programming.

I can’t write it directly, but after AI writes it, I can roughly understand it, then make fine adjustments. I can also describe my needs to AI quite clearly and have it modify further to meet my final requirements.

Where can this be applied?

For example, I’ve always advocated static hosting for landing pages, using lightweight and fast frameworks like Hugo/Astro, hosting them for free on Cloudflare Pages to replace the bloated, slow Wordpress that requires paid servers. This is because when I used Wordpress for blogging early on, a Wordpress plugin vulnerability led to an intrusion and privilege escalation, causing my site to redirect to random URLs. I definitely don’t want that drama happening while I’m scaling and sleeping.

But I don’t know how to write Hugo shortcuts or Astro components. How do I create the landing page website I want? Now with AI, these aren’t problems. Even though I’m not a frontend developer, with AI’s assistance, I can quickly build a Hugo/Astro-based website I want and host it on Cloudflare Pages to enjoy free, fast, and secure access.

Another example: normally, creating one campaign through the GUI is no problem and quite fast if you’re familiar with the process, but what about dozens of campaigns? Most affiliate networks, traffic sources, and tracking tools now have APIs. After clarifying the workflow, I can use AI to write a local tool, then use Cloudflare Workers to call the API, fetch offers from the network, batch add them to the tracker, batch create campaigns in the tracker, get campaign URLs, and then batch create campaigns in the traffic source. What used to take an afternoon manually and might have small errors can now be solved in minutes without mistakes by tooling this workflow.

There are many similar examples. After gradually tooling previously tedious daily workflows, efficiency has greatly improved and I’m less prone to errors. The next step is to figure out how to automate these tooled processes and become completely hands-off.

Insights and Tips

When I first started using AI, I always wanted to write a perfect prompt to have AI output the perfect answer I wanted, but in reality, perfection doesn’t exist. What we should do is start the conversation immediately, describe the needs clearly, provide enough information, and continuously supplement and polish through dialogue with AI to ultimately get the content we need.

When we get the final content, you can tell AI: Output the prompt needed to obtain the current information. Then learn from that prompt, and next time you ask questions, you’ll do better. We pursue excellence, but don’t pursue perfection, because perfection doesn’t exist. Done is better than perfect.

Summary

Whether AI is a bubble or not, I don’t know and I don’t care. I only know that AI has truly changed Media Buy, and changed my work and life. AI isn’t a silver bullet and isn’t omnipotent, but it is indeed a very good assistant.

If you haven’t started using AI yet, my advice is: start now. Don’t wait until you’ve mastered the perfect prompt to take action. Start with the simplest things like translation and copy generation, and gradually you’ll discover more application scenarios. As you continue dialoguing with AI, you’ll increasingly understand how to use it better.

For friends already using AI, I suggest establishing your own AI workflow. Review your daily repetitive tasks and see which can be AI-assisted, which can be tooled, and which can be automated. Every minute saved can be used to test more offers, optimize more campaigns, or simply enjoy life.

This industry changes fast, and AI is just one variable. But one thing is certain: the efficiency gap between Media Buyers who use AI and those who don’t will grow larger and larger. We don’t need to become AI experts, but at least we should make AI our capable assistant.