Article

Unlocking Hidden Value: A New Approach to Data Monetization with AI

Model Context Protocols are the key.

By Brock Simon, Bill Groves, Eric Sheng, Giorgio Saladino, and Vineeth Gogineni

First published on junho 10, 2025
min read


Many retailers and consumer products companies struggle with a recurring pattern: Organizations sit on mountains of valuable data that never gets used, let alone monetized. Not because the data lacks utility but because it lacks context. Most data monetization today still relies on flat files, APIs, or dashboards. These approaches assume the buyer knows what to ask for, understands how to use the data, remembers when to use it, and has the time to act on it. That’s a high bar, and one that leaves a lot of value on the table. But Model Context Protocols (MCPs) can provide a new way forward. Turning raw data into contextual answers MCPs allow AI systems to reason through complex questions by orchestrating roles, tools, data, and memory in a structured, context-aware flow. Instead of exposing raw data and hoping someone finds the insight, MCPs enable teams to deliver the answer directly—in context, when it’s needed, and shaped by best-practice logic. This changes what data is worth selling and how and where it’s used. More data becomes commercially viable. Many datasets that were too fragmented, niche, or context-dependent to monetize can now be activated within decision flows. By surfacing data only when it’s relevant to a specific question, MCPs remove the need to package and sell it up front. This makes monetization possible without the overhead of building a standalone product. Data can move beyond the head-office insight team. MCPs allow data to be used not just by strategy and insight teams at headquarters but across the front lines of decision making and action taking. Credit card sales data, geolocation trails, product reviews, and digital interaction histories, often used today in head-office planning, can now be brought directly into the workflows of store managers, digital merchandisers, marketers, and planners. That expands both the value of the data and the range of people it can support. Users don’t need awareness or deep understanding. If a decision or AI agent touches the data, MCPs allow it to be invoked behind the scenes with no discovery effort or manual activation required. Monetization becomes “just in time.” Instead of charging for access, organizations can monetize in-the-moment insights. The data is activated and paid for only when it directly supports a live decision. Best practices are embedded. With MCPs, organizations don’t just deliver data. They deliver how it should be interpreted and acted on. Logic, thresholds, and decision framing can all be encoded into the flow. Crucially, it also allows the businesses that own the data and understand its nuances best to embed their own expertise into the system. That means buyers get not just access to data but access to how to use it well, aligned to proven patterns and in-the-moment decisions. There is long-standing precedent for selling data in context. Credit scores are a prime example. They are widely monetized, deeply trusted, and specifically designed to support decisions about creditworthiness. They take complex, multidimensional data and collapse it into a single, actionable number. The same is true for other decision-driven metrics like Net Promoter Score^SM, customer lifetime value, ESG ratings, and brand equity indexes. Each is built to guide a specific class of decisions and has commercial value because of that tight alignment. But that’s also the trade-off: These models work for end users because they’re simple. They reduce complexity into a single number that’s easy to interpret and act on. But in doing so, they lose nuance, adaptability, and the ability to reflect real-time context. MCPs offer a different path. They allow us to preserve the richness of the underlying data, adapt to the specifics of the question being asked, and generate insight that evolves with the decision at hand. That opens the door to monetizing a much wider range of data assets without forcing them into a fixed, simplified format. We’ve heard skeptics say MCPs aren’t worth building. And in many internal use cases, they are right, especially when the system is narrow, the tooling is already optimized, or there’s no real need for multi-agent coordination. But that thinking assumes a single-organization view. When we zoom out across the ecosystem of retailers, manufacturers, data providers, and platforms, the value changes. At that scale, MCPs become more than a technical pattern. They become a connective layer: making data callable across boundaries, enabling shared reasoning, and turning siloed assets into decision-ready services. That’s where the cost makes sense and where the opportunity starts to look systemic. What still needs to be solved As promising as this model is, there are real challenges that need to be addressed before it can scale. Security. When data is invoked across tools or agents, strong safeguards must ensure it is only used by authorized parties. This includes protections against malicious access, data leakage, and misuse across systems. Governance. As MCPs scale, organizations will need central coordination points such as MCP gateways or mesh architectures. These manage permissions, route requests, and streamline cross-team data usage through a single, trusted entry point. Standardization and tooling. Many MCP implementations today rely on local developer setups that are not designed for secure or scalable use. Standardized MCP catalogs and toolkits, such as container-based deployments, can help organizations manage verified tools, streamline setup, and close critical gaps in discovery, credential handling, and server life cycle management. Performance and reliability. Agentic flows that depend on multiple calls, tools, or systems create more chances for something to fail. Building in fallback behavior, response time limits, and system resilience is essential. There is also an emerging need to guard against prompt manipulation, where malicious prompts hidden in documents or websites could mislead an AI agent. This requires layers of validation and content safety built into the design. Business model clarity. Even when data is integrated into workflows, the path to monetization is still unclear. Some companies have started charging based on individual tool calls. For example, remote MCP servers hosted on cloud platforms and SDKs like Stripe’s Agent Toolkit allow developers to wrap tool usage with pricing and access controls. These models offer early signs of how dynamic, usage-based monetization might work at scale. These aren’t reasons to avoid the opportunity, but they’re realities that need to be solved for. And the businesses that solve them early will have a significant head start. In our work with leadership teams, we’ve seen this firsthand. For most organizations, monetizing data isn’t just a product idea, it’s a business decision. It requires building a full commercial operation around the data: infrastructure, pricing, governance, support, and go-to-market strategy. That’s a big commitment, and many data owners won’t make the leap unless the opportunity is clearly massive. A smarter path to data monetization at scale What makes MCPs exciting is that they offer a lower-friction path. By embedding data into context-aware, question-driven flows, businesses don’t have to build a standalone data product from scratch. Instead, they can activate their data incrementally and participate in monetization only when that data is used to support a decision. Still, these are solvable problems. More importantly, they’re worth solving. Because if companies get this right, they can unlock entirely new forms of value by bringing dormant data to life, embedding insight in the flow of work, and reshaping how they think about decision making at scale.

Many retailers and consumer products companies struggle with a recurring pattern: Organizations sit on mountains of valuable data that never gets used, let alone monetized. Not because the data lacks utility but because it lacks context.

Most data monetization today still relies on flat files, APIs, or dashboards. These approaches assume the buyer knows what to ask for, understands how to use the data, remembers when to use it, and has the time to act on it. That’s a high bar, and one that leaves a lot of value on the table.

But Model Context Protocols (MCPs) can provide a new way forward.

Turning raw data into contextual answers

MCPs allow AI systems to reason through complex questions by orchestrating roles, tools, data, and memory in a structured, context-aware flow. Instead of exposing raw data and hoping someone finds the insight, MCPs enable teams to deliver the answer directly—in context, when it’s needed, and shaped by best-practice logic.

This changes what data is worth selling and how and where it’s used.

More data becomes commercially viable. Many datasets that were too fragmented, niche, or context-dependent to monetize can now be activated within decision flows. By surfacing data only when it’s relevant to a specific question, MCPs remove the need to package and sell it up front. This makes monetization possible without the overhead of building a standalone product.

Data can move beyond the head-office insight team. MCPs allow data to be used not just by strategy and insight teams at headquarters but across the front lines of decision making and action taking. Credit card sales data, geolocation trails, product reviews, and digital interaction histories, often used today in head-office planning, can now be brought directly into the workflows of store managers, digital merchandisers, marketers, and planners. That expands both the value of the data and the range of people it can support.

Users don’t need awareness or deep understanding. If a decision or AI agent touches the data, MCPs allow it to be invoked behind the scenes with no discovery effort or manual activation required.

Monetization becomes “just in time.” Instead of charging for access, organizations can monetize in-the-moment insights. The data is activated and paid for only when it directly supports a live decision.

Best practices are embedded. With MCPs, organizations don’t just deliver data. They deliver how it should be interpreted and acted on. Logic, thresholds, and decision framing can all be encoded into the flow. Crucially, it also allows the businesses that own the data and understand its nuances best to embed their own expertise into the system. That means buyers get not just access to data but access to how to use it well, aligned to proven patterns and in-the-moment decisions.

There is long-standing precedent for selling data in context. Credit scores are a prime example. They are widely monetized, deeply trusted, and specifically designed to support decisions about creditworthiness. They take complex, multidimensional data and collapse it into a single, actionable number. The same is true for other decision-driven metrics like Net Promoter Score^SM, customer lifetime value, ESG ratings, and brand equity indexes. Each is built to guide a specific class of decisions and has commercial value because of that tight alignment.

But that’s also the trade-off: These models work for end users because they’re simple. They reduce complexity into a single number that’s easy to interpret and act on. But in doing so, they lose nuance, adaptability, and the ability to reflect real-time context.

MCPs offer a different path. They allow us to preserve the richness of the underlying data, adapt to the specifics of the question being asked, and generate insight that evolves with the decision at hand. That opens the door to monetizing a much wider range of data assets without forcing them into a fixed, simplified format.

We’ve heard skeptics say MCPs aren’t worth building. And in many internal use cases, they are right, especially when the system is narrow, the tooling is already optimized, or there’s no real need for multi-agent coordination. But that thinking assumes a single-organization view.

When we zoom out across the ecosystem of retailers, manufacturers, data providers, and platforms, the value changes. At that scale, MCPs become more than a technical pattern. They become a connective layer: making data callable across boundaries, enabling shared reasoning, and turning siloed assets into decision-ready services. That’s where the cost makes sense and where the opportunity starts to look systemic.

What still needs to be solved

As promising as this model is, there are real challenges that need to be addressed before it can scale.

Security. When data is invoked across tools or agents, strong safeguards must ensure it is only used by authorized parties. This includes protections against malicious access, data leakage, and misuse across systems.

Governance. As MCPs scale, organizations will need central coordination points such as MCP gateways or mesh architectures. These manage permissions, route requests, and streamline cross-team data usage through a single, trusted entry point.

Standardization and tooling. Many MCP implementations today rely on local developer setups that are not designed for secure or scalable use. Standardized MCP catalogs and toolkits, such as container-based deployments, can help organizations manage verified tools, streamline setup, and close critical gaps in discovery, credential handling, and server life cycle management.

Performance and reliability. Agentic flows that depend on multiple calls, tools, or systems create more chances for something to fail. Building in fallback behavior, response time limits, and system resilience is essential. There is also an emerging need to guard against prompt manipulation, where malicious prompts hidden in documents or websites could mislead an AI agent. This requires layers of validation and content safety built into the design.

Business model clarity. Even when data is integrated into workflows, the path to monetization is still unclear. Some companies have started charging based on individual tool calls. For example, remote MCP servers hosted on cloud platforms and SDKs like Stripe’s Agent Toolkit allow developers to wrap tool usage with pricing and access controls. These models offer early signs of how dynamic, usage-based monetization might work at scale.

These aren’t reasons to avoid the opportunity, but they’re realities that need to be solved for. And the businesses that solve them early will have a significant head start.

In our work with leadership teams, we’ve seen this firsthand. For most organizations, monetizing data isn’t just a product idea, it’s a business decision. It requires building a full commercial operation around the data: infrastructure, pricing, governance, support, and go-to-market strategy. That’s a big commitment, and many data owners won’t make the leap unless the opportunity is clearly massive.

A smarter path to data monetization at scale

What makes MCPs exciting is that they offer a lower-friction path. By embedding data into context-aware, question-driven flows, businesses don’t have to build a standalone data product from scratch. Instead, they can activate their data incrementally and participate in monetization only when that data is used to support a decision.

Still, these are solvable problems. More importantly, they’re worth solving. Because if companies get this right, they can unlock entirely new forms of value by bringing dormant data to life, embedding insight in the flow of work, and reshaping how they think about decision making at scale.