Anastasia Braitsik stands at the forefront of the modern marketing landscape, navigating an era where data is simultaneously more abundant and more fragmented than ever before. As a global leader in SEO, content marketing, and data analytics, she has pioneered strategies that bridge the gap between technical tracking and high-level business intelligence. In a world where privacy shifts and device fragmentation have eroded traditional measurement, her expertise focuses on decoding the “signal loss” that plagues modern reporting. Today, she shares her perspective on why the quest for a single “source of truth” is often a trap and how marketers can find clarity amid the noise of modeled data and conflicting platform metrics.
Many reporting interfaces now blend observed outcomes with modeled estimates using similar visual styles. How can marketers distinguish between what is measured and what is reconstructed to avoid making confident but wrong decisions?
The danger in modern reporting isn’t that we lack data, but that our reports look far more precise than the reality behind them. When a platform like Google integrates machine-learning-based modeling into its measurement, it creates a seamless visual experience where a modeled conversion looks exactly like a directly observed one. My advice is to treat any metric labeled as “modeled” or “estimated” as directional rather than definitive, because these systems are essentially using patterns to guess what happened when the direct link is broken. If you lean too heavily on these reconstructions without questioning the underlying signal, you risk the practical consequence of budget misallocation at scale. You might see a report that suggests a channel is performing perfectly, yet your actual bank account or CRM tells a different story, leading you to confidently double down on a mirage.
You have spoken extensively about the “messiness” of the customer journey, where a single conversion might involve a podcast, a work laptop search, and a mobile retargeting ad. How should teams determine which touchpoints actually drive discovery versus those that are just the last identifiable interaction?
The user journey has become a series of disconnected fragments, and attribution systems often only see the parts they are able to connect. A person might hear about your brand on a podcast—an interaction that is almost entirely invisible to traditional tracking—then later perform a search on a laptop, read two articles, and finally convert after clicking a mobile retargeting ad. In this scenario, the retargeting interaction often receives disproportionate credit simply because it was the easiest touchpoint to measure, while the podcast and early research are classified as direct or organic. To combat this, I recommend that teams stop trying to find one “correct” model and instead triangulate across multiple methodologies. When you compare how a channel performs under different models, you start to see where the signals converge; if a channel’s contribution disappears the moment you move away from last-click, you know it was likely a closer rather than a discoverer.
It is incredibly common for teams to see Google Analytics 4 report 150 conversions, while Plausible claims 180 and the CRM shows only 120 new customers. When these “three realities” do not match, what is the most reliable way to anchor a marketing strategy?
This discrepancy doesn’t necessarily mean you have a data quality problem; it’s a reflection of different systems using unique attribution windows, conversion definitions, and reporting logic. One system might count every purchase, while your CRM only counts approved customers, and an ad platform might be using view-through attribution that the others ignore. I always suggest starting with the system closest to the actual business outcome, such as your backend order records or subscription data, to act as your “anchor.” Once you have that solid number—the 120 customers who actually paid you—you use the analytics and advertising platforms to understand the different parts of the journey surrounding those outcomes. This reframe changes the goal from finding the “right” number to identifying which touchpoints show up consistently across the journeys that actually resulted in revenue.
With the heavy push toward server-side tracking and first-party data collection, many believe the measurement problem is largely solved. How does this infrastructure change the actual quality of measurement, and what blind spots still keep you up at night?
First-party data collection and server-side tracking have become non-negotiable for anyone serious about measurement in 2026, but they aren’t a magic wand that recreates lost interactions. While server-side setups improve data reliability and give you more control over what is sent to platforms, they do not magically eliminate consent gaps or allow you to see interactions you were never permitted to observe. The value of this infrastructure is that it shifts the problem from a state of total ignorance to a “reasonable picture with known blind spots.” You gain a stronger first-hand observation of customer behavior, which makes you less dependent on external platforms to reconstruct your story. However, even with the best setup, you will still face gaps caused by cross-device transitions and strict privacy walls that no amount of technical wizardry can fully bypass.
What is your forecast for the future of attribution?
I believe we are moving toward a future where the job of attribution is no longer to provide “ground truth” but to act as one of several inputs designed to reduce uncertainty. Marketers will stop expecting their attribution stack to give them a single, perfect answer and will instead treat it as a directional guide that requires constant questioning. We will see a greater reliance on “triangulation” as a standard practice, where data from CRM, first-party tracking, and platform modeling are weighed against each other to inform high-level strategy. Ultimately, the marketers who thrive will be those who embrace incomplete answers and focus on making better decisions based on a “good enough” picture, rather than chasing the ghost of perfect precision in an unobservable world.
