Podcast Summary: Bonus Episode - The Marginal ROAS Revolution and How Meta's Algorithm Actually Works
Podcast Information:
- Title: DTC Podcast
- Host/Author: DTC Newsletter and Podcast
- Episode: Bonus: The Marginal ROAS Revolution and How Meta's Algorithm Actually Works with Constantine Yurevich of SegmentStream
- Release Date: April 16, 2025
Introduction
In this bonus episode of the DTC Podcast, host Eric Dick engages in an in-depth conversation with Constantine Yurevich, co-founder of SegmentStream. The discussion delves into the complexities of Return on Ad Spend (ROAS), the nuances between average and marginal ROAS, and the intricate workings of Meta's advertising algorithms. Constantine shares valuable insights on marketing measurement, attribution challenges, and effective budget allocation strategies for direct-to-consumer (DTC) ecommerce brands.
Understanding ROAS
Constantine Yurevich (A) kicks off the conversation by addressing common misconceptions surrounding ROAS. He emphasizes that ROAS is often used ambiguously, leading to confusion among marketers.
“[00:00] A: There are many misconceptions about ROAS and there are also many different things you call ROAS...”
Constantine explains that ROAS can be calculated in various ways—total revenue divided by total spend, profit divided by ad spend, or channel-specific ROAS based on different attribution models. This variability makes it challenging to compare ROAS metrics across platforms and campaigns.
Marginal ROAS vs. Average ROAS
The core focus of the episode is the distinction between average ROAS and marginal ROAS.
A illustrates this with a straightforward example:
“[03:11] A: Imagine you invest $1,000 a day in your Google search campaign and get $2,000 in return. Your ROAS is 2x...”
He contrasts this with a scenario where increasing the budget leads to diminishing returns:
“[07:25] A: Your average ROAS might still look healthy, but your marginal ROAS drops to 0.5, meaning each additional dollar spent only returns 50 cents.”
This differentiation is crucial because average ROAS can mask the inefficiencies in budget allocation, whereas marginal ROAS provides a clearer picture of the profitability of additional spend.
The Importance of Attribution
The conversation shifts to the challenges of marketing attribution. B (Eric Dick) recalls Constantine's previous views on the limitations of traditional attribution models.
“[00:55] B: ...you had some specifically hot takes around attribution being dead...”
A agrees, noting that many companies struggle to implement modern attribution models like Marketing Mix Modeling (MMM) effectively. He points out that as brands seek alternatives, technologies that better model attribution using data-driven approaches are gaining traction.
Budget Allocation Strategies
A significant portion of the discussion revolves around effective budget allocation. A warns against relying solely on average ROAS for decision-making:
“[08:50] A: The first fast win is to understand that when it comes to budget allocation, you should never look at average ROAS...”
He advocates for using marginal ROAS to guide budget allocations, ensuring that every additional dollar spent contributes positively to profitability. Constantine introduces the concept of controlled budget shifts, where budgets are dynamically adjusted to measure incremental impacts accurately.
The Role of Meta's Algorithm
A deep dive into how Meta's (formerly Facebook) algorithm functions follows. A explains the multi-layered machine learning models that Meta employs:
“[23:53] A: Meta has many machine learning models... account level, pixel level, and event level models...”
He highlights the importance of understanding these layers to optimize campaigns effectively. Constantine emphasizes that improper use of attribution and budget allocation can lead to suboptimal performance, regardless of the platform's sophistication.
SegmentStream's Evolution
B inquires about the evolution of SegmentStream’s platform. A describes its transformation from a visitor scoring attribution tool to a comprehensive marketing intelligence platform.
“[27:38] B: So that's a good sign. Let's talk a little bit more about SegmentStream...”
“[27:49] A: Yes. So at the moment SegmentStream evolved just from being a visitor scoring attribution and we talked about this a lot...”
The platform now offers modules for synthetic conversions, lead scoring, LTV scoring, and automated budget allocation based on marginal ROAS. This evolution aligns with the growing need for precise, data-driven marketing strategies.
Creative Strategies
The discussion briefly touches on the role of creative content in advertising campaigns. A advises against over-relying on constant creative testing within separate campaigns:
“[42:02] B: I have to ask you about creative...”
“[42:19] A: So frankly speaking, yeah, we're exactly on the data side...”
He recommends integrating new creatives into existing campaigns to leverage pre-trained models, ensuring that the creative elements align with effective targeting and optimization strategies.
Predictions for Media Buying
Looking ahead, A shares his perspectives on the future of media buying and the role of AI:
“[44:20] A: ...machine learning will definitely improve and automate a lot of processes but you should be always cautious...”
He predicts that while AI will continue to enhance marketing automation, the human element—particularly in setting accurate objectives and interpreting data—remains indispensable. Constantine warns of potential pitfalls if marketers do not align their objectives with the technology's capabilities.
Advice for Brands
Towards the end of the episode, A offers practical advice for DTC brands operating with budgets between $50,000 and $250,000:
“[34:02] B: ...what's the best way that they should think about getting started with marginal attribution...”
“[34:29] A: ...don't diversify too early... invest as much as possible into one channel, like Facebook, and master it before expanding...”
He stresses the importance of focusing on one channel to build robust machine learning models and avoid early diversification, which can dilute data and hinder effective optimization.
Conclusion
In closing, B encourages listeners to engage with SegmentStream for expert guidance on optimizing their media buying strategies.
“[46:54] B: ...if you want to get smarter about your media buying, you've got to go to segmentstream.com and talk to an expert...”
A reiterates the value of combining technology with expert consulting to navigate the complexities of modern marketing effectively.
The episode wraps up with mutual appreciation, highlighting the depth and practical value of the insights shared.
Notable Quotes:
-
Constantine Yurevich (A):
- “[00:00] There are many misconceptions about ROAS and there are also many different things you call ROAS...”
- “[03:11] Imagine you invest $1,000 a day in your Google search campaign and get $2,000 in return. Your ROAS is 2x...”
- “[08:50] The first fast win is to understand that when it comes to budget allocation, you should never look at average ROAS...”
- “[23:53] Meta has many machine learning models... account level, pixel level, and event level models...”
- “[34:29] Don't diversify too early... invest as much as possible into one channel, like Facebook, and master it before expanding...”
-
Eric Dick (B):
- “[00:55] Konstantin, welcome back to the DTC podcast... Let’s catch up on the world of attribution and marketing measurement.”
- “[08:17] It's such a moving target. Everything in e-commerce changes so dynamically...”
- “[34:02] What’s the best way that they should think about getting started with marginal attribution aside from giving you a call and subscribing to your newsletter.”
- “[44:53] Do you have any predictions for the media buying profession and the media buyers in the next two to three years?”
- “[46:54] If you want to get smarter about your media buying, you've got to go to segmentstream.com and talk to an expert.”
Key Takeaways:
- Differentiate ROAS Metrics: Understand the distinctions between average and marginal ROAS to make informed budget allocation decisions.
- Avoid Over-Diversification: Focus on mastering one advertising channel before expanding to others to ensure robust data for optimization.
- Leverage Technology with Expertise: Utilize platforms like SegmentStream for data-driven insights while incorporating expert consulting to maximize effectiveness.
- Understand Platform Algorithms: Gain a deep understanding of how platforms like Meta operate to optimize campaigns better and avoid common pitfalls.
- Strategic Budget Shifts: Implement controlled budget shifts to accurately measure and enhance marginal ROAS, ensuring sustainable profitability.
This comprehensive discussion provides DTC brands with actionable strategies to refine their marketing efforts, optimize ad spend, and navigate the evolving landscape of digital advertising with greater precision and effectiveness.