Milena Traikovich has spent her career at the intersection of performance analytics and high-impact lead generation, making her a crucial voice in today’s rapidly evolving marketing technology landscape. As companies scramble to integrate generative models and automated agents, Milena helps leadership teams look beyond the hype to find sustainable, measurable growth. In our conversation, we dig into the “ROI gap”—the strange reality where sales productivity rises by double digits while executive confidence in AI spending remains low. She shares her strategy for building robust data architectures and explains why the true cost of AI often hides in IT budgets or legal review cycles rather than the marketing stack itself. We also explore why the simple act of moving work between departments can masquerade as efficiency and how marketers can finally prove the financial worth of their technological investments.
The latest data shows sales productivity jumping by 14.1% and marketing overhead dropping by 14.6%, yet only a tiny fraction of CMOs can actually prove the ROI of these investments. Why is there such a massive disconnect between these impressive performance numbers and the financial clarity of AI programs?
We are seeing a fascinating friction between micro-efficiency and macro-profitability. While it’s true that customer satisfaction has climbed by 10.8% and individual tasks are being completed faster, most organizations lack the infrastructure to track these ripples across the entire ledger. Only 16% of CMOs feel confident in their measurement, largely because they are focused on the activity—the volume of content or the speed of a bot—rather than the specific financial change produced. When you consider that 68% of AI programs have gone over budget recently, it’s clear that the “savings” are being swallowed by unforeseen operational hurdles. To fix this, we have to stop treating AI as a magic wand and start treating it as a measurable business process that requires a defined baseline of performance from the very start.
You’ve often argued that data architecture is essentially the same thing as AI architecture in the modern enterprise. How does a company’s underlying data quality directly dictate the success or failure of their AI-driven lead generation efforts?
The hard part of finding value from AI is no longer about giving models access to more data; it’s about determining if that data can actually be trusted. By 2029, AI is expected to power more than 50% of all marketing activity in the U.S., but that power is useless if your systems are filled with duplicated records or stale information. An AI agent needs to know exactly which field takes precedence and what actions a specific piece of information permits before it can engage a lead. If your data architecture is messy, your AI will simply scale those inconsistencies, leading to conflicting signals that frustrate potential customers. We have to move toward a model where the data is clean enough for an agent to understand its meaning and its age in real-time.
Marketers frequently celebrate a 70% reduction in time for specific tasks like content creation, yet overall productivity sometimes stays flat or even declines. How does AI inadvertently move work instead of eliminating it, and what should teams be looking for to avoid this trap?
Generating hundreds of content variations in seconds doesn’t save much time if your employees then spend three hours checking them for brand compliance, duplication, and legal risk. We see this all the time: a 70% reduction in one task isn’t a real gain if the work simply moves from the “creator” to the “reviewer.” This is why it makes much more sense to measure the entire workflow rather than just the isolated task the AI performs. Sometimes, a less flashy tool that fits smoothly into how people actually work is far more valuable than a high-end agent that requires constant manual overrides. You have to account for the human hours spent fixing outputs, supplying missing context, or waiting for departmental approvals.
Calculating ROI becomes even more complicated when the costs and benefits are split across different departments, such as Marketing seeing the gains while IT or Legal absorbs the costs. How can leaders create a more holistic view of AI’s impact that accounts for these hidden expenses?
This is one of the biggest reasons AI ROI seems to disappear; the costs and benefits often live on different balance sheets. Marketing may get a productivity boost, but IT is the one paying for the infrastructure and engineering, while legal and security teams take on the burden of governance and monitoring. This can make a marketing initiative look incredibly profitable on paper because the heavy lifting is being subsidized by someone else’s budget. To get an honest number, you have to measure what changed across the entire company, not just within your own silo. You should only credit AI with financial results that you can reasonably connect to the specific work it changed, while counting every cost from integration to human review.
With nearly 70% of marketers admitting they cannot measure AI results with much precision, and 21% having no measurement infrastructure at all, where should a company start if they want to build a foundation for long-term success?
The first step is to stop measuring the activity AI performs and start measuring the change it produces in the bottom line. You need to establish exactly how a process performed before the AI was deployed so you have a legitimate baseline for comparison. Without this, the more removed a benefit is from the activity, the harder it becomes to prove that the AI actually caused it. It is also vital to build a consistent infrastructure for measurement now, rather than waiting for the technology to mature further. If you don’t have a way to track the integration, data preparation, and training costs today, you’ll never be able to justify the increased spending we expect to see over the next three years.
What is your forecast for the state of marketing AI?
As we look toward 2029, the industry will shift from a period of wild experimentation to one of rigorous architectural discipline. We will see the “ROI gap” close as companies stop chasing every new feature and instead focus on tools that integrate seamlessly into their existing human workflows. The winners won’t be the companies with the most agents, but the ones with the cleanest data and the most transparent measurement systems. I expect that the 50% of marketing activity powered by AI will eventually be managed by a new class of “operational auditors” who ensure these systems stay within budget and brand guidelines. Ultimately, the focus will move away from what AI can do and toward what AI can actually prove it has earned.
