Tackling the data management challenge: Insights from Enterprise Connect

This week at Enterprise Connect, I had the privilege of joining industry leaders to discuss one of the most pressing challenges facing businesses today: effectively managing and leveraging dispersed data stores to improve customer experience. As someone who has spent the past 15+ years building data platform solutions across the BPO (Business Process Outsourcing) and tech industries, I've witnessed firsthand how the proliferation of data sources has created both opportunities and obstacles for delivering exceptional customer experiences.

Our panel, moderated by Zeus Kerravala (Principal Analyst, ZK Research), brought together experts from across the communications and customer experience space to explore how businesses can transform fragmented data into actionable customer intelligence. I'd like to share some key insights from our discussion.

The myth of a single source of truth

One of the first topics we addressed was whether contact centers should be the central consolidation point for all customer experience data. In my view, this approach fundamentally misses the mark.

A customer's experience with a brand isn't just shaped by the contact center, but by every interaction across different functions of the business. Searching for a single system to hold all CX data is a fallacy; no one platform can capture everything that happens with a customer.

Instead of pursuing this elusive "single source of truth," I believe businesses should focus on creating a dynamic customer experience data layer that connects insights from all touchpoints into a unified profile. This approach ensures that different departments can access relevant customer information in the most meaningful context.

What makes a truly effective unified profile?

For virtual and human agents at contact centers to deliver personalized experiences, they need to accurately derive intent in a timely manner and access contextual data immediately. AI is powerful only when such contextual data is available to it. Effective unified profiles must provide:

•  Live signals from different systems: For example, visibility into products customers are browsing, items in abandoned carts, recent in-store purchases, and real-time shipping updates
• Relevant transactional details: For example, seamless access to order history, appointments, payments, and account updates without agents needing to dig through multiple tools
• Cross-channel communication history: For example, context about previous interactions, marketing offers, and outstanding issues to help AI agents avoid unnecessary escalations—and, when escalation is necessary, ensuring human agents have everything they need without asking repetitive questions, avoiding transfers, or keeping customers waiting

Most importantly, every customer interaction should enrich the profile with summaries, intent analysis, and sentiment data, creating a continuous improvement cycle for more relevant and personalized experiences.

The distributed data reality

When our panel addressed whether businesses should aggregate data into a unified set or leave it distributed across systems, I advocated for a more nuanced approach.

A strong warehouse strategy is essential, but the goal is not to centralize everything. It's about making the data usable, in the right place, at the right time.

At Twilio, we've observed that data engineers typically prefer source data to remain in place, allowing them to run AI experiments closest to the data while still enabling activation in various business tools. Meanwhile, business teams demand flexibility in their data stack, wanting the ability to replace old applications as new channels and tools emerge.

The problem isn't that data is distributed, it's that most businesses lack easier ways to unify and activate it in real time at scale. The solution isn't moving all data into one place, but rather, it involves creating effective connections between disparate sources within the right customer context.

Our approach to data interoperability

During the panel, I shared how Twilio is planning to address these challenges by solving both communications and customer data interoperability:

• Advanced identity resolution: We combine probabilistic and deterministic matching to clean up messy data and eliminate duplicates
• Schema-less engagement event APIs: Our solutions connect unified profiles with hundreds of data sources for real-time activation 
• A semantic layer for data graphs: We link warehouse and CRM data with zero/low-copy architecture 
• Native integration with communication solutions: Our platform simplifies the process of gathering engagement insights

Beyond APIs: The need for new standards

While APIs help move data between systems, they don't solve the fundamental challenge of making that data meaningful. Today's APIs require developers to manually integrate systems, learn different formats, and maintain these connections over time.

With AI-driven automation, we expect AI agents to take over system integrations, reading metadata, dynamically discovering APIs, and adapting to schema changes. For example, an AI handling product returns should automatically understand that "order_number" and "transaction_id" refer to the same thing, without human intervention.

However, AI needs guardrails. Just as prompt engineering is challenging, AI agents must not only communicate effectively but also take appropriate actions. This suggests the need for new standards to govern AI-driven integrations, ensuring that AI agents act responsibly, securely, and within intended limits.

The future of AI agents

Looking ahead, our panel explored whether businesses will converge on a single AI agent or develop multiple specialized agents. My belief is that most organizations will need a network of interoperable AI agents with different specializations.

Just as companies don't use one coding language for everything, they won't rely on a single AI agent for all tasks. Businesses will need a network of interoperable, general-purpose AI agents and those specializing in different domains.

For example, a general-purpose support AI might need to pass insights to a sales AI when identifying a frustrated customer as a churn risk, or coordinate with a fraud AI that specializes in detecting suspicious behavior. Forcing a single AI agent to handle all these scenarios would be inefficient and lack the precision needed for each use case.

Also, much like human agents, AI agents can't operate in silos. The need for shared customer context is pervasive across every function. A support AI helping a customer with a billing issue needs to know if a sales AI recently offered them a promotional discount, so it doesn't mistakenly escalate a non-issue. Without a unified view of customer data within the context they’re handling, these AI systems will make disjointed, misinformed decisions, leading to poor customer experiences and inefficiencies.

Securing data across systems

Our panel concluded by addressing the critical issue of security across disparate data sources. The challenge extends beyond securing data at rest to include securing data in motion, ensuring compliance, and maintaining customer trust across every interaction.

Consumers expect businesses to honor their data privacy choices and notification preferences. Conversely, businesses must ensure  strict compliance by safeguarding their data from bad actors, obtaining consent before using it to train AI models, and adhering to strict regulatory requirements.

At Twilio, security and privacy aren't afterthoughts; they're baked into every product we build. We anonymize sensitive data, enforce encryption, apply strict access controls, use advanced identity and AI powered fraud solutions, and honor consumer privacy preferences when ingesting and sharing data across systems.

Moving forward

As businesses continue to navigate the complex landscape of data management in the age of AI, I believe the path forward involves embracing distributed data realities and preparing for a future of specialized AI agents. By moving beyond the myth of a single source of truth, organizations can transform their approach to customer data management.

The ultimate goal remains consistent: creating seamless, personalized customer experiences that build trust and loyalty. With the right strategies for data unification, security, and activation, businesses can turn the challenge of fragmented data into an opportunity for differentiation through superior customer engagement.