Milena Traikovich is a seasoned leader in the demand generation space, known for her ability to transform complex data into high-performance lead-nurturing engines. As an expert in performance optimization, she has spent years navigating the evolving landscape of ad platforms and marketing automation, helping businesses bridge the gap between technical execution and strategic growth. In our discussion, we explore the critical necessity of moving beyond vague objectives to define clear operational boundaries for AI. We dive into the hidden risks of efficiency, the importance of establishing “playing fields” rather than just directions, and why the human element is more vital than ever in an era where systems can autonomously chase the wrong goals.
When an automated system is tasked with maximizing a specific metric like return on ad spend, what are the common ways it might actually work against a company’s long-term growth interests?
The reality is that automation doesn’t possess the business intuition to distinguish between high-quality growth and hollow efficiency. If you hand a system a vague goal like “maximize ROAS,” it will naturally find the path of least resistance to hit that number, which often means leaning heavily into branded search and retargeting people who were already going to buy. You might see your ROAS climb significantly on the dashboard, but you aren’t actually acquiring new customers; you’re just paying for the ones you already had. This creates a feedback loop where the system feels successful because the metric improves, but the actual incremental value to the business is stagnant. It’s a classic case of the automation doing exactly what you asked for while completely missing what the business actually needed to thrive.
You’ve suggested that marketers should stop giving automation a simple direction and start giving it a “playing field.” How does this shift in mindset change the way a campaign is structured?
Giving a system a direction, such as “we need higher ROAS,” is an open-ended invitation for an optimizer to chase that number right off a cliff. Instead, you have to define the sidelines and the floor—the exact conditions under which a win is still considered a win for the company. For example, if a brand is currently running campaigns at an 8x ROAS but wants to prioritize growth, the leadership must explicitly decide how much efficiency they are willing to trade for volume. You might tell the system that you will accept a drop from an 8x ROAS down to a 5x ROAS, provided that the volume of new customer acquisitions grows in proportion. By setting that 5x floor, you give the AI the room to move into less efficient territory and experiment with broader audiences while ensuring it doesn’t spend its way into a deficit.
How does this challenge of vague objectives manifest within CRM workflows and lead nurturing programs beyond the world of paid media?
It is incredibly easy to get caught up in the technical excitement of building elaborate automated workflows in a CRM, but many of these are just engineering efforts layered on top of untested assumptions. We often trigger emails, assign sales tasks, and nudge customers toward specific actions without any data proving that these actions actually lead to better retention or higher lifetime value. If a human wouldn’t perform that task manually because the value is unclear, then automating it simply means your guesses are now running on a set schedule. You might end up filling your funnel with a high volume of low-intent signups that look great in a weekly report but never actually activate or generate revenue. This results in a “noisy” system where the automation is working perfectly, yet the business isn’t seeing any tangible progress.
In highly regulated sectors like insurance, what are the specific dangers of adopting a “default” posture when using advanced tools like Google’s AI Max?
For an advertiser in a regulated industry, the default setting of “enable everything” is a recipe for a compliance nightmare that no dashboard will ever catch. You might find the AI rewriting your carefully vetted ad copy into something that sounds punchy but hasn’t been approved by your legal team, or it might pull brand terms into broad matching categories where they simply don’t belong. The most effective move in these scenarios is to decide exactly what to turn off before you ever turn the system on. By disabling text customization or excluding specific brand terms, you aren’t showing a lack of trust in the technology; rather, you are giving the system a safe field where it can sprint without crossing legal or ethical boundaries. Guardrails are the only thing that make full autonomy safe enough for a professional marketing team to actually deploy at scale.
As automation takes over the tactical heavy lifting of bidding and targeting, how does the role of the human marketer need to evolve to remain effective?
The human job is no longer about approving every minor bid change or manually reviewing every suggested action; it has shifted to owning the high-level definition of the field. We are now the ones responsible for setting the floors, defining the exclusions, and identifying the specific tradeoffs the business is willing to accept to reach its next milestone. Human judgment earns its keep by watching for those critical moments when a system is winning the metric but losing the game. We have to be the ones to step in and say that a “win” doesn’t count if it comes at the expense of brand integrity or long-term customer value. If every automated system you run hits its target this quarter, you still have to ask yourself if those wins actually moved the needle for the business or if they were just mathematical successes in a vacuum.
What is your forecast for the future of automated marketing objectives?
I believe we are heading toward a period of “corrective maturity” where the industry moves away from chasing raw volume and back toward nuanced, business-aligned constraints. We will see a shift where the most successful marketers are those who spend 20% of their time on execution and 80% on defining the “loss conditions” and guardrails for their AI agents. As ad platforms become even more autonomous, the competitive advantage will lie in who can most accurately translate complex business goals into the “playing fields” that these systems require. We will stop seeing AI as a “set it and forget it” solution and start treating it as a high-performance engine that requires a very specific track to stay on course. Ultimately, the winners will be those who realize that automation is a multiplier—if you multiply a vague guess, you just get a faster failure, but if you multiply a well-defined strategy, you achieve true scale.
