AI Revolution in Paid Social Marketing: Strategies, Trends, and Insights with Mike Taylor

Introduction

In today's fast-moving marketing world, AI is a game-changer. It's bringing fresh and exciting ways to help teams boost their strategies and create new opportunities.

Think about it: AI can help make catchy ad copy, assist with brainstorming, and even come up with creative testing concepts. For those in paid social marketing, AI isn’t just nice to have—it’s becoming a must-have tool.

We talked to Mike Taylor, an expert in both the advertising and AI worlds.  Mike co-founded Ladder, a standout marketing agency with a 50-person strong team, but that’s not all! Mike is also leading Vexpower, a startup with a clear mission: helping marketers get savvy with data and tech skills.

For Mike, it’s all about training a new generation of experts who can blend classic marketing know-how with a sharp technical edge. His LinkedIn courses have helped over 300,000 students fine-tune their skills in ad tech and AI. Plus, Mike has been one of Kitchn.io's earliest supporters, showcasing his dedication to innovative marketing tools and platforms!

Recently, Mike has plunged into the fascinating world of prompt engineering (and even created a Udemy course!), where he's exploring new and powerful ways to use AI. In our chat with Mike, we'll delve into the real-world ways AI is changing how paid social teams build and run their campaigns.

The Impact of AI on Paid Social Marketing

Why did you decide to shift your focus toward AI?

After spending six to seven years growing my agency, I began to realize the toll it took in terms of long hours and relentless client requests. Anyone familiar with the life of a media buyer understands the demanding nature of the job. The hustle was real, involving a lot of laborious, brute-force work just to keep clients satisfied.

During my years at Ladder, I developed an interest in the more technical parts of advertising. I managed to grasp the basics of coding and even dived into ad automation. In fact, the reason I met Simon from Kitchn is that he was the only other person that asked about the API at a Facebook event in Dublin. When I eventually left Ladder, my primary goal was to embark on a more technically enriching journey. My first three months post-Ladder were dedicated to a data science boot camp where I familiarized myself with Python programming.

But, of course, it didn't stop there. I was eager to explore more. Three distinct pathways captured my attention: tracking and analytics, marketing mix modeling, and generative AI. First, there was the complex world of tracking, where many businesses stumble. The evolving complexity, particularly after the iOS 14 changes, represented a considerable challenge.

Next, I immersed myself in marketing mix modeling—a statistical technique capable of operating independently of tracking. I saw it as a form of insurance given the unpredictability of the tracking landscape post-iOS 14.

And finally, there was AI. I had been involved in several AI projects with our data science team and was fascinated, but when generative AI came out it felt like magic. In 2020, GPT-3 beta was live, and I showed my wife, a copywriter at the time. Her reaction told me this was something that was already good enough to cause serious disruption, and I needed to dive deeper to get ahead.

At the time, when I was writing my blog posts, I would spend weeks diving into economics textbooks and exploring marketing mix modeling or be interpreting dense privacy legislation and browser policies in order to determine what will happen with tracking. I'd put a ton of effort into a single blog post, and maybe it would get 100 or 200 visits, tops.

But with AI, it’s a whole different story. I whipped up a blog post in an afternoon on prompt engineering, and suddenly, it’s racking up 10,000 more visits than anything else I had written. That was a wake-up call. It must have felt to be part of the Internet's early days.

Interestingly, up until recently, the bulk of my consulting revenue was derived from marketing mix modeling, tracking, and analytics. It was a significant change when AI started generating a more substantial share of my earnings. The tipping point for this transition was AI consistently surpassing the expectations of clients. I realized that despite all the hype, AI had the potential to yield results that far exceeded my projections, even with minimal input. The promise and thrill of AI's capabilities fueled my decision to make it my focus.

I was always jealous I wasn’t around during Silicon Valley's golden age, watching the internet take shape and witnessing the birth of groundbreaking startups. It was scary to turn down lucrative consulting contracts for a new field, but, I decided to commit fully to AI. I realigned my entire portfolio towards AI projects, dropping anything that I couldn’t pivot towards AI.

How is AI currently shaping the landscape of paid social marketing?

It’s really important to break down AI into categories, especially when it comes to image and text generation. When we talk about image generation, one standout use is ad creative production and creative testing.

I’d say that with some of the latest models, like Midjourney, the creatives are already at a point where they're good enough to move straight into production with certain clients. Of course, it really depends on the client—different clients have their own unique styles or standards they’re aiming for. But here's the thing: if your client isn't currently impressed with the state of the art, just give it six months. I'm willing to bet they'll be blown away by then.

Now, with AI integrating smoothly with tools like Photoshop, Figma, and others, the process of creating visuals has become a whole lot simpler. Plus, AI’s knack for tweaking images opens up fresh, inventive routes—for instance, it can pull out and transform specific elements of an image in ways we might not have thought of.

Where I believe AI is really essential is during the prototyping phase. It lifts a ton of mental weight off our shoulders when it comes to coming up with new creative concepts. These AI-crafted prototypes are must-haves for communication—they help us get our ideas across to both our teams and our clients, speeding up the approval process. And when it's time for a brainstorming session, the ad angles that AI can generate might breathe new life into our ideation process.

Balancing AI and Human Creativity

How can AI ensure ad quality compared to working with a creative team?

To make sure AI's integration doesn’t lower the quality of our ads, we need a strategy. The role AI plays in prototyping and brainstorming is just huge. It’s a powerhouse at churning out a wide variety of ad concepts, and that’s a game-changer. It’s my starting line for fine-tuning the creative work. With AI, the brain-draining task of coming up with ideas from zero is eased. Plus it seriously cuts down the time I spend on that initial burst of concept creation.

But let’s be clear: AI isn't here to replace human creativity. You shouldn't look at AI as doing the whole task. You should really look at it as removing steps from that task or decreasing the amount of time that task takes because it's going to get you really close really quickly. But you might still need that human touch to get it exactly perfect for production.

I haven't heard of anyone actually letting their design team go. Even heavy adopters of AI aren’t doing that. What they are doing, though, is pumping out creative assets like there’s no tomorrow. We’re talking ten times the amount, easily, or they’re just exploring way more ideas than before. Think about it: before, you’d maybe get four or five new designs every couple of weeks. But now? It could be more like 50. That means you can just pick the best ones and get them ready for testing. It's like suddenly having this massive sandbox to play around in.

AI is a star player when it comes to kickstarting creative development and handling the repetitive stuff, but when it comes to striking that perfect balance between creativity and efficiency, nothing beats the human touch.

Besides generating headlines and ad copy, are there other AI applications in text generation?

Without a doubt, AI is great at more than just writing headlines and ad text.

One key use is organizing product feeds better. Most of the time, those feeds are very messed up, sometimes even scraped from a website. They might pull in errors or even bits of HTML code.

Other times, the headline is just “white blanket”, or “white mattress” — super basic, no flair, essentially just a listing. But often, there's a goldmine of information hidden in the product descriptions or attributes that could totally be leveraged in the ads. It’s just that the data isn’t neat or consistent. For instance, some products might clearly state the color, while others leave it out—although it's right there in the description.

So, a big part of what I've been collaborating on with agencies is cleaning up those product feeds. Imagine breezing through 1,000 products in the feed in just a few minutes with AI. It sifts through the data, plucks out the key attributes, and seamlessly incorporates them into the titles. What you're left with is a much crisper, cleaner product feed that’s polished and ready for your ads.

And this becomes ultra-important when the ads are being auto-generated—think dynamic product ads. These are the sorts of tasks that, without AI, could suck up your entire day. And let’s be honest: sitting there writing 1,000 headlines means a horrible day.

AI is also really good at analyzing data and doing research. Let’s paint a picture: you pull down 100 competitor ads, snapping screenshots or scraping the text, all in the hope of spotting some trends.

Now, ask yourself: do you really want to be the person painstakingly combing through every single one of those 100 ads, mentally sorting them into categories—like, ‘which ones focus on price,’ ‘which ones highlight location,’ ‘which ones are all about speed’?

Ideally, you’d tag these ads with these attributes to spot patterns, like ‘Ah, 50% are emphasizing speed, but only 10% are playing up the price.’ That intel helps you decide your angle. Should I stand out by talking about the price since almost no one else is, or should I jump on the speed bandwagon since it’s clearly a popular approach?

AI can quickly go through a huge number of competitor ads, spotting trends and categorizing details. This helps to reveal what strategies and trends competitors are using, which can guide new ideas and decisions.

The Power of Prompt Engineering in AI

How can paid social teams benefit from prompt engineering? What do they need to look at to get good at?

Prompt engineering is so powerful that it can greatly improve how paid social teams use AI, and this is currently a main focus of a book I am writing for O'Reilly Media. The key idea here is that the way you ask AI to do something can greatly affect the quality and usefulness of the results. It's interesting that the same question can lead to different answers from AI, and this can be a problem when the answers are not what you want.

Prompt engineering is all about optimizing how you ask AI to do something to get more reliable and relevant results. In my experience, I've found several techniques that make AI communication work more like effective human communication.

Adding examples of past tasks

For example, adding examples of past tasks to your prompts can make them more reliable. If you show the AI that you want "speed" to be a theme in an ad, it can learn to use that theme consistently. Structured formatting is also important, especially in technical settings like Kitchn.io, as it helps with smooth data transfer.

Giving clear directions

Being clear and direct with your instructions to the AI, like you would be when instructing a person, helps guide the AI toward specific tasks. It's also helpful to include style guidelines in your prompts to keep the brand’s voice and identity consistent.

Test, Test, Test

Testing is key. Running a lot of tests on important prompts you plan to use in production helps identify what needs to be improved. By tracking how often issues come up and tweaking the prompts accordingly, you can make them much more reliable. This is similar to A B testing in marketing, which is used to find the best-performing option.

Watch the prompt length

Being mindful of the length of your prompts is smart, as longer prompts can increase costs without necessarily improving results.

Despite how advanced AI models have become, the role of prompt engineering remains critical. Human guidance is still needed to get the most effective results from AI.

Interestingly, prompts are starting to play a role similar to that of top-notch ad templates. As Andrej Karpathy, the former AI head at Tesla, wisely noted, this process is like a new kind of programming that uses natural language instead of traditional code.

Essential AI Tools for Paid Social Teams

What are the AI tools that are the most valuable and the most useful in everyday life for paid social teams? Which tools would you recommend using and looking at?

Without a doubt, Kitchn.io is a key tool for adding AI to paid social teams. It’s great at organizing data and smoothly incorporating prompts into work processes, which brings big benefits.

Other tools, like Supermetrics and Funnel.io, are really helpful for collecting ad spending data in one place, like a data warehouse, and connecting it with AI platforms later on. Zapier is also a big help because it makes work processes run more smoothly.

While there are many AI tools to choose from, I've found that working directly with ChatGPT, often through its API, usually gives fantastic results without costing too much. OpenAI is the biggest player in the world of AI trends and new developments, so it’s definitely one to keep an eye on.

For teams that aren't very technical, tools like Jasper and Copy AI are great options. They come with set prompts and custom models, so there’s less need for detailed prompt engineering.

For more technical teams, though, I generally suggest getting comfortable with using AI tools directly. This approach helps teams understand the tools better and allows them to adapt as AI technology changes. Langchain is emerging as a strong tool for managing AI prompts and making the process more efficient.

Looking forward, steady progress in image generation is exciting, with models like Midjourney and LLMOps set to play key roles.

AI's Impact on paid social job market

What roles do you see disappearing through the usage of AI in paid social teams?

I like this saying: “AI isn't going to take your job, but someone using AI will.” AI is changing how people work in paid social teams, and this brings up some interesting points to think about. As AI tools and automation keep getting better, we can expect some traditional jobs to change. For example, tasks that people used to do by hand, like optimizing bids, are increasingly being done by automated systems. People who don't start using AI might have a harder time finding jobs or keeping up with the changing standards of the industry.

However, it’s unlikely that AI will completely replace human workers. Instead, as AI becomes more common, new jobs are likely to be created that focus on managing, tweaking, and monitoring AI-driven processes. Over time, we might see people doing fewer manual tasks and more work that involves strategy, creativity, and making decisions.

As things change and some of these tasks that you're doing manually right now get automated, you don't have to do all this manual work anymore. You can do more of a strategy.

You can spend more time working with finance to convince them of the value of marketing. You can spend more time coming up with interesting creative strategies that can't easily be done by AI. You can spend more time coming up with new channels to test or new tactics to try. I think overall, the job just gets better and more interesting over time as these boring things get automated.

Take manual bidding as an example. It used to be a common task in platforms like Google Ads and Facebook, but now automated systems are doing much of this work. People who are willing to learn about AI and improve their skills will be in a good position to help shape the changing landscape of the industry.

Outlook to the future: emerging AI trends

Considering your extensive experience and the rise of AI in marketing, what emerging trends should paid social teams be aware of to maintain a competitive edge?

Given the dynamic AI landscape, paid social teams should stay attuned to several emerging trends:

Agent-Based AI

We’re seeing more AI systems that can learn and improve on their own. They work in a cycle—setting tasks, making responses, checking the results, and adjusting their processes by themselves. This is bringing a new level of independence to AI.

Advanced Image Generation

Tools like ControlNet and Segment Anything are changing how images are edited. Segment Anything is great for making precise masks, and Control Net gives detailed control over image creation. These tools are opening up new possibilities for creative work.

Testing and Debugging AI Prompts

As AI becomes more important, it will be crucial to test and check prompts thoroughly. A B testing, which involves comparing two versions to see which one performs better, is likely to become a regular part of this process.

Hybrid AI Approaches

We might see more people using a variety of AI models and tools together. This approach would let teams combine the strengths of different AI systems to get specific results.

Consistency and Aesthetics

With more advanced tools for creating images, keeping a consistent and attractive visual style, especially in videos, will become a key concern.

Skill Enhancement and Adaptation

Keeping up with new developments in AI will be essential. Teams that are willing to embrace AI and learn about its growing capabilities will be in a strong position as the landscape changes.

While AI adoption rates vary, staying informed about these trends can position paid social teams for success in an AI-driven future.

Conclusion

The impact of AI on marketing is huge. We hope that we gave you an insight into how AI and human skills can work together. AI helps in many areas of marketing, but it doesn't replace the need for human creativity. It's like using a new tool to make your job easier.

As we look ahead, marketers need to learn how to use AI while keeping their unique touch. So, as we move forward, remember that combining AI with our own skills will lead to great results in marketing.