Anastasia Braitsik is a titan in the digital ecosystem, specifically known for her ability to decode the complex relationship between data architecture and marketing performance. As a global leader in SEO, content marketing, and data analytics, she has spent years navigating the shifting tides of how brands connect with their audiences through technology. Currently, she is at the forefront of a major shift in the industry: the transition from marketing technology as a collection of specialized tools to martech as a foundational infrastructure decision. With the emergence of “data gravity” as the primary force shaping modern stacks, Anastasia provides a critical lens on why the traditional methods of vendor consolidation are failing and how a universal data layer is becoming the only viable path for AI-driven growth.
Our discussion explores the evolution of Customer Data Platforms (CDPs) into warehouse-native entities and the emergence of data gravity as the defining metric for organizational efficiency. We delve into the sobering statistics regarding CDP utilization, the strategic choice between platformization and agentification, and why marketing teams must prioritize zero-copy access to ensure AI readiness. Anastasia also breaks down the architectural choices marketers face today, highlighting the importance of governance and data integrity in creating a consistent customer experience across fragmented touchpoints.
Large, fragmented datasets often create “data gravity” that slows down AI decisioning and increases operational costs. How are you seeing this shift the way marketers view their technology stacks?
Data gravity is no longer just an abstract concept; it is the physical drag that occurs when customer, campaign, and intent data are trapped in separate systems with completely different schemas. In this year, we have moved past the era of the “audit spreadsheet” where martech consolidation simply meant cutting tools that no one logged into. Instead, we are seeing that every new tool tends to pull its own copy of data rather than reading from a shared warehouse, which creates silos that make integration unmanageable. As these datasets grow, the cost of moving data becomes prohibitive, and every AI initiative built on top of this fragmented stack inherits that same heavy lag. Practitioners are beginning to realize that the immediate trigger for consolidation might be rising costs, but the true driver is the need for data integrity to power autonomous decisioning. When the governance burden becomes visible to leadership, the conversation shifts from a simple cost-cutting exercise to a deep infrastructure overhaul aimed at eliminating the drag of fragmented information.
With only 22% of marketers reporting high utilization of their CDPs, why is there such a disconnect between the investment and the actual value realized?
The disconnect stems from a “plumbing” problem where teams have spent years adding new “fixtures” like specialized CDPs, reverse ETL tools, and identity graphs without considering how the pipes actually connect behind the walls. It is a staggering reality that while 41% of companies have implemented a CDP, organizations estimate they only use roughly 47% of the capabilities they are actually paying for. This underutilization happens because many packaged CDPs centralize data inside a proprietary database, essentially duplicating information that already lives in the enterprise warehouse. This duplication leads to “low water pressure” in the data flow, where marketing teams are forced into constant reconciliation meetings to figure out why their CDP record doesn’t match their source of truth. Until the stack moves away from these redundant systems that require constant manual syncing, marketers will continue to struggle to find the value in tools that are essentially anchored by their own complexity.
Databricks recently entered the marketing space with CustomerLake. How does this shift the definition of a CDP from a marketing tool to a piece of core enterprise infrastructure?
The launch of CustomerLake this past June marked a definitive signal that customer data management is shifting away from the marketing application layer and closer to enterprise data platforms. By building an agentic CDP natively inside a lakehouse environment, Databricks has effectively challenged the very existence of standalone CDP vendors. This move forces CMOs to treat their next renewal as a fundamental infrastructure decision rather than a typical software purchase, as the data warehouse is now directly capable of marketing execution. We are seeing a push toward consolidating around fewer, more integrated platforms because maintaining a consistent customer data layer is becoming nearly impossible in a proliferated tool environment. This shift simplifies governance and improves interoperability, creating a much stronger foundation for AI, which we know performs best on structured, consistent, and well-governed data rather than fragmented copies.
You’ve mentioned that the CDP market is splitting into platformization and agentification. Could you elaborate on what these paths mean for the daily workflow of a marketing team?
The split in the market is essentially a choice between two different ways of handling the data layer: platformization anchors the CDP within a broader suite from vendors like Adobe or Salesforce, while agentification keeps the CDP “thin” and hands execution to autonomous agents. For the average marketer, this shift should ideally manifest as a reduction in “swivel-chair” reporting, where they no longer have to jump between five different tools to get a single view of a campaign’s performance. When a consent preference is updated in a properly consolidated data layer, that change is pushed across every touchpoint at once, ensuring that a shopper who opts out of tracking on an app isn’t immediately retargeted with stale data on a social platform. The workflow win is directly tied to the customer experience win because both rely on having one governed version of the customer record. By moving toward a universal data layer, we are removing the friction of manual data movement, allowing marketers to focus on strategy rather than spending their mornings reconciling disparate spreadsheets.
For teams looking to modernize, how does the choice between a packaged CDP and a warehouse-native composable layer impact their AI readiness?
AI readiness is fundamentally a question of data duplication; the more copies of a customer record you have, the higher the chance that your AI will be making decisions based on “hallucinated” or outdated data. A warehouse-native, composable architecture scores the highest on AI readiness because it utilizes zero-copy access, meaning the AI agents are querying the governed source of truth directly. In contrast, point-to-point integrations create high duplication and inconsistent governance, which leaves the AI with fragmented context that can lead to disastrous customer-facing errors. Even a packaged CDP, while more centralized, often depends on the coverage of its connectors, which can create gaps in the data the AI is able to “see.” By enforcing governance at the source—the warehouse—marketers ensure that any autonomous agent querying the platform for decisioning is using the most accurate, structured, and permissioned data available.
Before a major renewal, what specific experiments should a marketing leader run to validate their data architecture choice?
I strongly advise marketers to treat their next renewal as an infrastructure test and to bring their IT or data teams into the conversation as early as possible. A powerful way to validate a new path is to pick a single, contained channel—such as an abandoned cart email flow or a loyalty tier update—and run it through a warehouse-native activation path for a full quarter. You should compare the maintenance load and the freshness of the data against your existing “packaged” tools to see which model actually reduces the operational drag on your team. During these evaluations, you must ask vendors point-blank if their platform requires a separate copy of your data or if it can read directly from your warehouse using zero-copy access. If a system cannot show its reasoning or trace an AI-driven recommendation back to the source data, it simply isn’t ready to handle unsupervised, customer-facing decisions in the current landscape.
What is your forecast for CDP consolidation?
I expect that by 2030, the standalone, siloed CDP will be a relic of the past, with 80% of net-new deployments being either embedded in or fully composable with enterprise data platforms. We are moving toward a reality where the application layer finally answers to the data layer, reversing a decade-long trend of building “fixtures” that dictated how our data should be structured. This means the primary metric for a successful stack will no longer be how many vendors you have, but how many places a single customer record must live before an agent can act on it correctly. As zero-copy architecture becomes the standard, the “data gravity” that once slowed us down will become our greatest asset, providing a dense, unified core of information that allows AI to drive personalization at a speed and scale we are only just beginning to realize. The winners of the next five years will be the organizations that stop moving their data and start moving their applications to where the data already lives.
