Unlocking AI-Powered Media Strategies

**(00:00)**  

We have a fun conversation planned on media buying with Abtin Buergari from Blueprint. Abtin was introduced by our community member, Chris Perkins, and it’s wonderful to see how speakers are emerging from our own network. Abtin, thank you for joining us. I’ll make you a co-host, so feel free to share your screen if needed. Just a reminder to everyone: this is an interactive conversation, not a webinar. Feel free to jump in, raise your hand, or add thoughts in the chat.

**(02:05)**  

Thanks for inviting me! I’m Abtin Buergari, CEO of Blueprint, an AI tech product for the ad tech industry. A bit about my background: I’m a transplant to marketing.

I previously worked in the legal field, where I built a company focused on helping attorneys quickly locate relevant evidence among millions of records by structuring unstructured data. After selling that company, I started looking into new ventures and began investing in e-commerce businesses around 2017. I wanted to evaluate engagement and understand how e-commerce businesses stack up against each other, but there weren’t good tools available for cross-comparing advertising. 

This sparked the idea for Blueprint. First, I created an agency—now run by Chris Perkins—and started managing large ad spends, which helped me better understand the messiness of advertising data. We were managing over $100 million in ad spend at the agency, and by 2021, I began building the foundation for Blueprint, which launched in 2023. We’re now scaling rapidly, managing $350 million in ad spend through our platform.

**(05:28)**  

Great background, Abtin! A big question I have is if we’re heading toward an “easy button” for advertising—where a business could upload some basic content and information, select a few goals like driving online sales or building an email list, and let AI handle the rest. Could we be heading toward that?

**(08:00)**  

That's a great analogy. It’s similar to Tesla’s robo-taxi vision—an ideal that will take time to achieve due to complexities we can’t fully anticipate. I think we’ll eventually reach a point where AI can handle the details, but there’s still a lot to solve. To get there, we’ll need to address complex issues like data transparency and platform dynamics.

**(09:07)**  

I’d love to hear your thoughts on Web3 and blockchain’s role in addressing some advertising challenges. Blockchain offers an open ledger, which could potentially address data transparency and privacy issues by allowing only limited access to data without it being stored by platforms.

**(09:47)**  

Yes, blockchain presents an interesting solution for privacy, allowing data to be stored securely without platforms like Facebook fully accessing it. But platforms are reluctant to embrace blockchain solutions because they prefer advertisers to stay within their ecosystems. So while it’s technically feasible, the practical adoption of blockchain by these companies remains uncertain.

**(12:20)**  

When I first entered advertising as an engineer, I was struck by the chaotic way data is organized. Many agencies present outdated data to clients without meaningful insights, but brands that take control of their data often outperform agencies. I’m seeing more brands build internal media teams, sometimes just 10-12 people, but with incredibly tight control and oversight of their ad spend.

**(15:25)**  

A smart marketing team accepts that there will be some data loss and uses a system that their executives can trust, like Shopify for e-commerce or HubSpot for B2B. These platforms should be the “source of truth” for ad performance, instead of relying on incomplete data from multiple sources.

**(17:06)**  

David, to your earlier question about the “easy button,” I think we’ll get closer once we automate the mechanical aspects of advertising. The strategy and content choices, however, need to stay human-driven. Machines can crunch data, but determining the right audience and message remains a human responsibility. For example, while ChatGPT can answer many questions, it won’t know if a website accurately represents an audience’s needs.

**(19:09)**  

Right now, some companies aren’t benefiting as they should because they aren’t coordinating their SEO, paid ads, and organic strategies effectively. Agencies often manage different pieces in isolation, which means clients don’t see the full impact or benefit.

**(21:20)**  

To build on that, Abtin, what do the latest advancements in AI-powered media buying offer that wasn’t possible a year or two ago?

**(21:55)**  

Good question. There’s some exciting work happening now. One example is technology that pulls video ad content directly from platforms like Facebook, analyzes engagement data, and uses large language models to extract insights about what’s in the video—who’s speaking, the message, etc.—and suggests what other content might engage viewers. One company doing this is Darwin, but this technology is still evolving and mostly a point solution.

Our platform, however, focuses on ad fatigue and growth potential. We analyze when ads start to fatigue and notify users so they can make changes before they spend too much on underperforming content. We also identify high-performing content and send it to an AI model to analyze characteristics, though we’re still refining how best to use this data to help clients make actionable decisions.

**(26:28)**  

Karan, you raised an important point. Clients often want to keep content running longer than is effective because they’re attached to it. Our technology helps identify when content engagement starts to decline quickly—what we call “fatigue”—and alerts clients so they can make adjustments.

**(29:37)**  

Jay-Z also asked about optimizing across channels since every platform’s dynamics are different. Our platform tracks the same creative across platforms and shows how it performs in each one. This helps clients make data-driven decisions about where to use their best content.

**(33:36)**  

Our platform flags high-performing creatives and shows where they’re most effective, allowing users to optimize content based on specific platform insights. This helps media teams spend less time creating generic reports and more time focusing on real-time decisions, such as reallocating budgets or creating similar content.

**(38:59)**  

Karen, about large language models (LLMs) and share of voice, that’s a fascinating area. We’re seeing LLMs create inherent biases based on available content, which impacts share of voice in search and conversational AI results. While our platform doesn’t currently focus on this, I think there’s real potential for using LLMs to continually adjust content for different platforms based on user engagement.

**(41:21)**  

Amit, great question on reinforced bias. When you invest heavily in specific creatives, the risk is that the algorithm may overemphasize one piece of content without exploring new options. At Blueprint, we address this by not relying solely on return on investment (ROI) metrics. Instead, we look at ad performance across different dimensions—engagement rate, conversion rate, and the entire funnel. We’ve found that analyzing ad sets and comparing platform data with attribution systems like GA4 provides a more complete picture.

**(47:12)**  

Jay-Z, on your question about whether our platform can auto-optimize like some “multi-armed bandit” models—yes, there are tools like metadata.io that do this for B2B, but we’ve chosen not to pursue this approach. Automation can be risky if something goes wrong or if the data isn’t entirely accurate. Instead, we focus on providing insights so media operators can make educated decisions based on our recommendations.

**(52:09)**  

Thanks, Abtin, for the incredible insights and for sharing your experience. Thank you all for the fantastic questions and engagement. We have several more speakers lined up in the coming weeks. Looking forward to seeing you all next week. Abtin, we’d love to have you back again—thanks for joining us!