Enterprises have poured millions into AI, yet personalization often falls flat not because the models are weak, but because the data feeding them is broken. Fragmented customer records scattered across channels and business units leave AI unable to tell whether it’s looking at one customer or five, undermining every recommendation, campaign, and cross-sell opportunity built on top of them. In this conversation, Ankit Utreja, Co-founder and CTO of WebEngage, unpacks why the decades-old user-event model has become a structural liability, what relationship-based intelligence unlocks for complex customer hierarchies like banking, and how CIOs can build governance frameworks that keep pace with India’s evolving DPDP Act all while making the business case that turns modernization into measurable revenue.
Co-founder and CTO
WebEngage
CIO&Leader: Enterprises have invested heavily in AI, yet personalization outcomes remain underwhelming. What is the single biggest architectural reason AI initiatives underperform — and why does it trace back to customer data?
Ankit Utreja: For clarification, by AI, we mostly mean Generative AI powered by LLMs. In that light, Personalization is not solely an outcome of AI; rather, hyper-personalization at large scale relies heavily on machine learning (ML) rather than Generative AI, based on the end user’s past trail and intent. And for machine learning to work, data sanity needs to be maintained by Enterprises. Many enterprises have fragmented customer databases, with records distributed across channels, products, and business units. This causes fragmented, incomplete, or disconnected signals; this makes it very difficult for the models to identify intent, context, and relationships of customers accurately. Intelligence depends on the quality of the underlying data foundation; if an individual is stored differently across systems, the model will think there are multiple individuals when there is just one. This will result in inconsistent recommendations, irrelevant customer engagement, and missed opportunities for the organization. The real challenge is creating a unified, continuously updated, and comprehensive view of the customer, taking into consideration both their behaviors and relationships. When customer data can be connected and contextualized, AI/ML can understand customers’ behaviors and relationships, enabling it to move from just predicting outcomes to delivering truly personalized experiences at scale.
CIO&Leader: At what point does the decades-old user-event model become a structural liability, and what are the early warning signs that a data architecture is actively constraining AI ambitions?
Ankit Utreja: Organizations moving from campaign automation to AI-led customer engagement are increasingly finding the user-event model a hindrance rather than an asset because it cannot capture a customer’s broader context. Indicators that organizations are experiencing these problems include inconsistency in personalized marketing across multiple channels, increasing difficulty establishing a single customer for multiple products and accounts, increasing reliance on manual segmentation, and an increasing amount of time and effort required to create actionable insights.
Additionally, organizations are experiencing AI models that provide accurate technical prediction ratings but do not translate into business results. As customer journeys become more complex, enterprises require architectures that capture both what customers do and their relationship to the product. Without that context, the ability to use AI will eventually hit a ceiling.
CIO&Leader: For a CIO managing complex customer hierarchies — a bank customer holding a loan, savings account, and credit card — what does the shift to relationship-based intelligence unlock that event-based models simply cannot?
Ankit Utreja: Relationship-based intelligence enables organizations to view customers as a web of related entities rather than as individual events across various systems. When considering the banking industry, a customer can have a savings account, a credit card, a loan, an investment account, and multiple family members. Traditional event-based systems can monitor customer activity across all these products, but have difficulty understanding how each event affects the other events. A relationship-based architecture enables organizations to aggregate multiple customer touchpoints, giving AI a much better view of overall spending patterns and life-stage needs, thereby enabling enhanced cross-sell opportunities, proactive service delivery, and better risk mitigation. Rather than just responding to isolated events, organizations will be able to view the entire customer relationship as a cohesive whole and deliver more meaningful engagement. The result will be to strengthen customer loyalty, increase retention, and enhance organizations’ ability to deliver individualized experiences throughout the full life cycle of the customer relationship.
CIO&Leader: When modernizing customer data architecture, what criteria should a CIO use to decide whether to extend an existing CDP or data warehouse or adopt a purpose-built platform?
Ankit Utreja: When making this choice, the business’s outcome should take precedence over preferred technologies. The CIO must determine whether their current design can provide a single view of every customer in real time, manage complex relationships with multiple customers, and accommodate additional uses of AI in the business. Traditional data warehouses work best for storing and analyzing data, while many CDPs primarily provide features for joining customer profiles across various channels.
However, as personalization and AI become top priorities, organizations require new systems that can consistently execute, improve, and utilize customer data across multiple channels. As businesses grow, the source of truth for their data naturally becomes their own data stores. No matter how much a third-party system claims to be a source of truth, it is only true for a function or two in the organization. Composable customer data platforms enable CIOs to invest in their own data lakes and data warehouses, own the truth, and still provide all the functionality of a CDP on top of them.
CIO&Leader: How does a relationship-aware data layer integrate with existing CRMs, ERPs, and data lakes — without replacing them? What does a realistic enterprise integration roadmap look like?
Ankit Utreja: The relationship-aware layer of your application is intended to complement existing systems, not replace them. For example, your CRMs will continue to serve as your system of record for customer interactions; enterprise resource planning (ERP) solutions will continue to provide operational management. All of these systems are in the cloud or hosted on-premises. Then you will layer the relationship-aware aspect on top of them to integrate customer identity, product relationships, behavioral signals, and business context into a single intelligence framework.
A realistic roadmap for implementing a relationship-aware data layer begins by integrating the most valuable customer data sources and establishing a consistent customer identity layer across them. The second phase of your roadmap will focus on creating customer profile integrations and enabling real-time customer data flows across various channels, continuing from phase one.
Finally, over time, your organization will be able to introduce AI-driven use cases through a phased approach that minimizes disruption, maximizes current technology investments, and provides a means to demonstrate business value early while building an intelligent customer ecosystem.
CIO&Leader: As AI takes on automated decision-making over sensitive customer data, how should CIOs structure governance frameworks — especially given India’s evolving DPDP Act obligations?
Ankit Utreja: Governance will no longer be an exercise in compliance; it needs to be incorporated into the AI strategy as a key component. As AI usage continues to increase in customer engagement and decision-making processes, CIOs require an appropriate governance framework that balances innovation with accountability. Effective frameworks begin with a strong consent management system, clear and transparent data-use policies, and strict access and processing controls. Evolving data privacy regulations require organizations to maintain a clear record of how they collect, store, and use customer data. Furthermore, organizations must ensure all AI-based decisions can be explained, audited, and aligned with business policies. Governance must be embedded directly into the architecture of customer data, enabling ongoing monitoring and compliance by design. Organizations that treat trust as a competitive differentiator will be better positioned to leverage AI effectively, maintain customer trust, and remain prepared for regulatory requirements.
CIO&Leader: Do architectural principles for relationship-aware data translate across BFSI, retail, travel, and healthcare — or must each sector build a fundamentally different foundation for AI effectiveness?
Ankit Utreja: Even though all industries are largely built upon the same set of foundational principles, industry differentiation is primarily found in how these foundational principles are manifested in the form of existing customer relationships. For example, in banking, credit cards and bank account relationships typically are involved; while in retail, customers typically have a relationship with their purchases as well as their preferences and their loyalty programs; travel companies generally want to understand customer relationships as they relate to their bookings, their destinations and their service interactions; and healthcare organizations have relationships with patients, providers and the treatment provided to the patient over time. While every industry will have its own unique use cases, there is one commonality that all organizations can agree upon: To provide positive outcomes, AI must be able to access relevant, comprehensively connected customer data. Organizations that build flexible architectures that support modeling complex customer relationships can be responsive to industry-specific needs while maintaining a consistent, shared platform for providing personalized experiences, retaining customers, and intelligently engaging with them.
CIO&Leader: Modernizing data infrastructure is a significant commitment. How should a CIO build the board-level business case — what metrics demonstrate that this investment translates into measurable retention or revenue outcomes?
Ankit Utreja: The strongest business rationale for modernizing customer data is the direct relationship between modernization and measurable growth results for the company. Boards seem far less interested in how the technology will be improved than in its impact on revenue, customer retention, and operational efficiency. As such, CIOs should create specific metrics for customer lifetime value, repeat purchase rates, churn reduction, engagement growth, conversion rate improvements, and the effectiveness of marketing campaigns. CIOs should also quantify the negative effects on the company from not having a unified view of the customer, such as lost sales opportunities due to “missed” cross-sell opportunities, poor experience for customers, and the cost of carrying out processes manually to resolve the issues caused by having fragmented customer data. The use of modern customer data architectures enables companies to engage with their customers more effectively and respond more quickly to changes in customer behavior.
CIO&Leader: What infrastructure trade-offs must CIOs navigate to deliver genuine real-time AI personalization at enterprise scale, without compromising data consistency or system resilience?
Ankit Utreja: To achieve real-time personalization, you need a balance between speed, precision, and scalability. Many businesses prioritize low latency but fail to consider the need for a consistent and reliable view of their customers. The real challenge is processing large volumes of customer data in real time while ensuring that the decisions produced are accurate across all channels. CIOs must design their architectures to enable continuous data ingestion, identity resolution, and activation without delays. Equally as critical to success is building resilience, observability, and governance into your architecture design from the very beginning. A successful architecture will enable you to achieve both speed and reliability by creating a mechanism that realizes both simultaneously. The goal is not simply to be faster but to be smarter in how you respond. There is no value in providing real-time personalization unless a business can trust the data being used for every customer interaction.
CIO&Leader: If you were advising a CIO beginning a customer data modernization program today, what would you prioritize in the first 90 days — and what is the one investment most organizations defer that costs them the most?
Ankit Utreja: The focus for the first 90 days is to deeply understand where the current customer data is located within the organization and to identify which business outcomes can be realized with the greatest impact. Many modernization efforts fail because they are approached as a “Technology Project” rather than an “Enterprise Transformation Project.” The investment most organizations put off is building a solid foundation for customer identity and relationships. Many organizations spend time on dashboards, analytics, or AI models before addressing the fundamental problem of compiling all information about each customer to form a Connected Customer view. This shortcut approach limits long-term success. By establishing a strong identity layer early in the process, an organization is laying the foundation for every AI, personalization, and customer engagement initiative it undertakes. A strong identity layer is one of the highest return-on-investments an organization can make.