Independent Agency Global Overview on Powering Streaming TV Ads by Pairing AI & Human Storytelling

Ari Paparo (0:10)

Welcome to Marketecture, where you can get smart fast with in-depth interviews of leading executives. I'm Ari Paparo. I'm here at the Amazon Unboxed Conference in San Francisco. We're here to talk about a pretty interesting subject, which is using agents to create video ads and enabling streaming TV. I have two guests today. So we have Claire Magruder, who is the Director of GO Studio at the agency called Global Overview, along with Santosh Gunaseelan, who is the Director of Amazon Ads for Creative Intelligence and Optimization.

Claire and Santosh, thank you so much for being here.

Claire Magruder (0:43)

Thank you. Thanks for having us.

Ari Paparo (0:45)

So, start, Claire, tell us about Global Overview and what we're going to be talking about today.

Claire Magruder (0:50)

Yeah. Global Overview is a full service agency. So, we help brands perform on Amazon and really kind of optimize their business in any way that fits their goals and objectives. We're focused on ads, but we also offer creative support and we drive retail performance. We have operations. We have the full spectrum of support at Global Overview.

Ari Paparo (1:10)

Got it. And Santosh, tell us about what you've been working on and what brings you here.

Santosh Gunaseelan (1:14)

Yeah. I had Creative Intelligence Optimization, which is a team that creates these suite of tools to produce creative for campaigns. But we also layer in a deep intelligence to ensure this is not only just good-looking ads, but also perform well.

Ari Paparo (1:28)

Yeah. It's incredibly important because with AI, generative AI, everyone feels like they can just create some video ads and run them, but they have to actually be good.

Santosh Gunaseelan (1:36)

Yeah. Now, that's right. I mean, one of the things today with frontier models is, I mean, you can go and try frontier models, anything for that matter, and you'd get a good piece of content. And I think there's a massive conflation between good content and good ad. And we want to be at the latter end of it. We want to be able to use good ads.

Ari Paparo (1:52)

It's not always obvious what's good. So let's talk about a case study that you folks have come out with that seems pretty interesting. So the brand is Traditional Medicinals. So Claire, what is Traditional Medicinals?

Claire Magruder (2:03)

Yeah. They are an established company that produces tea. And it's herbal tea, and it's all about wellness. And so they have a wide range of products, and it really is meant to support people and their daily health goals from a more holistic, herbal perspective.

Ari Paparo (2:19)

And so there are active advertisers, but not so much with streaming.

Claire Magruder (2:23)

That's right. Yes. So they were doing well on Amazon with Search and Display, and they've been running the business there. But they really knew they needed to get in front of more customers. The new to brand piece is so key. And they knew that they perform well, especially with women 45 and up.

And so the goal was, how can we go find these people and help educate them on the actual, the brand and who they are?

Ari Paparo (2:49)

There's a whole tier of advertisers who don't participate in the sight, sound, and motion of video and streaming, right?

Claire Magruder (2:55)

Right.

Ari Paparo (2:56)

And Santosh, that's a big part of why you're investing here, right?

Santosh Gunaseelan (2:58)

I think frontier models have come a long way. However, what they don't account for today is, you could go to any frontier models out there. We use a lot of them. We'll be super model agnostic. They will not adhere to brand guidelines.

They would not know what's your outcome that you're actually targeting and what's your objective for, what kind of audience are you trying to reach. And this is where we are focusing on. Because the more general outputs create what we call as this AI slop today. That's what you see in the market a lot. It really throws off audiences. And so we've been investing heavily in this, what we call as this intelligence layer. And it factors in brand guidelines. It knows exactly who you're trying to target and what's the message that needs to be there. And that becomes part of this creative brief and that gets evaluated against multiple signals. And we find out that if this is most likely going to perform. And then we go generate this.

And you can think of this as a quality, a creative quality and a performance harness, not just a generation engine.

Ari Paparo (3:57)

Yeah. And for those listening at home, I guarantee you this podcast is not AI Slop. We are real human beings, really in the Bay Area, really talking for now.