Governance First: The Key to Scalable, Trusted Data

Executive Summary

Why enterprise analytics needs governance at the core

As AI-powered analytics moves from pilot projects to enterprise-wide adoption, organizations face five persistent challenges: inconsistent answers, security and compliance risk, vendor lock-in, rising costs, and growing complexity. These problems stem from a single root cause: inadequate governance.

The 2025 Global AI+BI Survey highlights this governance gap. While most organizations are using advanced analytics, only 45.5% report enterprise-wide governance, and just 20% enable natural language access. This fragmentation erodes trust and stalls growth.

Solving this requires a universal semantic layer: a governance-first foundation independent of any single tool, cloud, or database. It decouples governance from tools, platforms, and data sources, ensuring consistent logic and secure access across every layer of the modern analytics stack.

This whitepaper introduces a governance-first approach to data and analytics—powered by Strategy Mosaic, a universal semantic layer for enterprise analytics.

Strategy Mosaic is a universal semantic layer that lives independently of your analytics tools, databases, and clouds. It provides the consistency and trust required for enterprise AI, from reducing hallucinations to enabling cost arbitrage.

Strategy Mosaic turns complex, federated environments into cohesive, trusted systems—supporting not just dashboards, but also AI agents, customer-facing apps, and future use cases still to come.

This whitepaper explores the five core challenges holding AI-powered analytics back and how Strategy Mosaic solves them with a single, governance-first foundation.

Introduction

AI-powered analytics is moving from isolated pilots to enterprise-wide adoption. Organizations no longer want siloed tools; they want AI that delivers real business value across departments and teams.

But scaling is not just a technical challenge. It is, fundamentally, a governance challenge.

The 2025 global survey confirms this governance gap:

• Only 45.5% of organizations have implemented enterprise-wide governance.

• Just 20% allow employees to interact with data using natural language.

• This is in contrast to the 58.7% that use advanced analytics and the 52.3% that have standardized their data.

^{Source: State of AI+BI Analytics Global 2025 Survey}

This means the majority of organizations lack the trust infrastructure to scale AI safely and consistently, as a shared layer for business definitions and access controls is needed to move AI beyond departmental silos and expert users.

Saurabh Abhyankar, Chief Product Officer at Strategy, draws a distinction between three types of enterprise intelligence:

• Type 1: Departmental insights powered by individual tools like Excel or Tableau. These are valuable but siloed and fragmented.

• Type 2: Enterprise-wide intelligence built on governed, semantic, and connected layers that unify logic and access across tools, clouds, and teams.

• Type 3: Autonomous intelligence, where AI not only augments decisions but proactively analyzes data, surfaces insights, and acts within business context.

The shift from Type 1 to Type 2 is well underway. But the leap to Type 3—autonomous, contextual, and embedded AI—requires one thing above all: governance that scales.

This level of control must be independent of any single tool, database, or cloud to ensure the flexibility and consistency needed for true enterprise AI. This is where a universal semantic layer comes in, providing a trusted, flexible foundation for scaling analytics.

That's what Strategy Mosaic delivers. By decoupling governance from the underlying infrastructure, Mosaic provides the consistency and trust required for enterprise AI.

Problem #1: Inaccurate and Inconsistent Answers

In our 2025 global survey of 235 organizations, less than a third of respondents had full confidence in the accuracy and reliability of AI-powered analytics.

That lack of trust isn't irrational; inaccurate or inconsistent answers, so-called "hallucinations," were the most-cited technical challenge—reported by 43.4% of organizations. These inconsistencies don't just slow down decision-making; they erode trust and often require time-consuming quality assurance before results can be acted on. As one survey respondent puts it:

Survey respondent:

"Most things need to be quality assured. This is still time-consuming, as QA feeding back into the AI model in an automated way is still a challenge."

Top Challenges in AI-Powered Analytics (Figure 1)

Survey question: From technical and operational perspectives, which of the following present problems for your AI-powered analytics? (Select all that apply)

^{Source: Dúnedain Research, The State of AI+BI Analytics Global 2025 Report, Sponsored by Strategy}

How Mosaic Solves Inaccuracy

Strategy Mosaic solves this problem with a universal semantic layer that acts as the single source of truth for all business logic.

When a team uses a data analytics tool, an AI agent, or a custom application, they are all querying the same definitions for metrics like revenue, inventory, and product. The logic doesn't live inside those tools; it lives in Mosaic's central semantic layer. This ensures that a single metric is always calculated the same way, regardless of the user, tool, or query.

Saurabh Abhyankar, Chief Product Officer, Strategy:

"Mosaic provides a single set of models with clearly defined business definitions—like revenue and product, for example. Whether you're accessing them from Tableau, Excel, or through an AI tool using SQL, Mosaic calculates it the same way and gives you the same answer across all platforms."

By establishing one central source for all business definitions, Mosaic:

• Lowers the risk of hallucinations by feeding models a governed, trusted semantic model, not raw, unmanaged data.

• Eliminates contradictory answers that cause confusion and a lack of trust.

• Reduces the need for manual QA because the logic is already certified and consistent.

This approach is made faster and more reliable by Strategy Mosaic's AI-powered modeling studio. Mosaic handles the initial, time-consuming work of building the semantic model. What used to take days of manual labor can now be done in minutes, allowing a human expert to focus on fine-tuning and validation. This automation:

• Reduces human error: With less manual work, there's a lower risk of mistakes that lead to data inconsistencies.

• Accelerates time to trust: What used to take days can now be done in minutes, allowing teams to get a consistent, reliable model in place faster than ever before.

The result is a cohesive analytics stack, where every query is built on a foundation of trust, consistency, and accuracy.

Problem #2: Security and Compliance

Security policies embedded in siloed tools don't travel across platforms or clouds, leading to exposure or inconsistent enforcement.

According to our 2025 global survey, 52.0% of organizations cite regulatory risk, including data privacy, bias in AI models, and sector-specific rules as their top adoption challenge.

At the same time, most organizations have not yet built the internal scaffolding to ensure that AI systems only access what they're authorized to see.

With increasing pressure to scale AI access to non-technical users, companies face a key dilemma: How do we empower teams and AI tools without putting sensitive data at risk?

Top Challenges to Adoption (Figure 2)

Survey question: Which of these challenges has impacted or may impact your adoption of AI-powered analytics across the organization? (Select all that apply)

^{Source: Dúnedain Research, The State of AI+BI Analytics Global 2025 Report, Sponsored by Strategy}

How Mosaic Solves Compliance Issues

Mosaic allows organizations to enforce consistent security and compliance policies, without needing to duplicate rules across platforms.

Mosaic decouples governance and security from underlying tooling, providing:

• Row-level security that defines access by region, department, or product line.

• Column-level security that restricts sensitive fields like salary or personal identifiers.

• Feature-level security that governs what users can do with data—whether exploring, exporting, embedding, or querying with AI.

Saurabh Abhyankar, Chief Product Officer, Strategy:

"One user might only be allowed to view data from a particular region, while another sees different segments. Mosaic allows you to define very sophisticated security filters at every level."

When applied to AI tools, this governance model becomes especially impactful.

Saurabh Abhyankar, Chief Product Officer, Strategy:

"You can expose data to AI tools in a tightly controlled way—ensuring that the AI only sees the data it's authorized to work with. That way, your AI initiatives remain compliant, secure, and aligned with business rules from day one."

This governance is not layered on, it is built into the semantic foundation. Mosaic eliminates the need to duplicate policies across systems. Security travels with the logic, by design.

Mosaic turns governance from a constraint into an enabler, allowing organizations to scale AI confidently, even in highly regulated environments.

Problem #3: Portability and Vendor Lock-In

Enterprise analytics environments today are highly diverse. According to the 2025 global survey, organizations typically use:

• 3 analytics platforms (Power BI, Tableau, Strategy One, Excel, Looker)

• 2 cloud providers (Azure, AWS, Google Cloud)

• A variety of structured and unstructured data platforms

Many vendors still assume a closed-stack model, trapping business logic in a specific tool or a single database. This becomes a major problem when priorities shift due to cost, performance, compliance, or major changes to the business.

PeggySue Werthessen, VP of Product, Strategy:

"You need to be able to react, leverage new technologies, and respond quickly to events such as mergers and acquisitions. This is where our independence comes to bear. We are not beholden to any particular stack. Our mission is to help you use any technology to manage costs and risks for your entire data estate."

How Mosaic Solves Vendor Lock-In

By decoupling governance and business logic from any single vendor's stack, Mosaic enables organizations to maintain flexibility without compromising consistency.

This separation is critical to avoiding vendor lock-in. When semantic definitions are embedded inside one database or tool, migrating platforms becomes complex and costly.

Saurabh Abhyankar, Chief Product Officer, Strategy:

"If your semantic layer is embedded in your database, you're stuck. You'd have to move your data, rebuild the layer, and then update everything on top of it. It's a massive undertaking."

Mosaic prevents that scenario by being truly open. It allows you to:

• Keep your data in any warehouse: Whether in Snowflake, Databricks, BigQuery, or Hadoop.

• Query from any tool: Use Power BI, Tableau, Excel, or custom APIs.

• Operate across any major cloud: Switch providers as your infrastructure strategy changes.

Saurabh Abhyankar, Chief Product Officer, Strategy:

"Mosaic isn't just dual open—it's open in three critical ways: cloud, data sources, and consumption. That's what makes it so flexible and interoperable."

This architecture transforms vendor diversity from a limitation into an advantage. With Mosaic, you can:

• Select the best tool for each team or use case.

• Migrate between platforms without duplicating logic.

• Future-proof your infrastructure as technology landscapes evolve.

In a world that demands speed and adaptability, Mosaic gives your organization the freedom to adapt—without rework, risk, or disruption.

Problem #4: Cost Arbitrage

AI-powered analytics is often pitched as efficient and transformative, but without cost control, it becomes unsustainable. Unexpected compute bills, duplicate modeling, and rigid pricing models continue to undermine the business case for scaling AI-powered analytics.

In our 2025 global survey:

• 49.0% of organizations cited cost, especially lack of predictable ROI, as a major barrier to adoption.

• 46.4% listed cost savings as their top goal for the next 12–18 months.

Survey respondent:

"Our finance team dreads reviewing monthly cloud bills. Surprises are frequent—and expensive."

AI-Powered Analytics Goals: Next 12–18 Months (Figure 3)

Survey question: What are your organization's most important AI-powered analytics goals for the next 12 to 18 months? (Select all that apply)

^{Source: Dúnedain Research, The State of AI+BI Analytics Global 2025 Report, Sponsored by Strategy}

How Mosaic Enables Cost Arbitrage

This allows teams to shift workloads, reduce redundancy, and contain compute costs without sacrificing performance or control. The true cost of scaling isn't just about software licenses: it's about the time wasted on rebuilding logic, moving data, and over-consuming compute power. Mosaic addresses these drivers head-on by enabling cost arbitrage.

Saurabh Abhyankar, Chief Product Officer, Strategy:

"One of the things we specifically designed Mosaic to do is help you achieve cost arbitrage."

Mosaic reduces the total cost of ownership by:

• Enabling cloud and database flexibility: Teams can shift workloads to the most cost-effective platforms based on their needs.

• Preventing duplication: Mosaic connects to your data wherever it lives, eliminating the need for expensive ETL (extract, transform and load) or data consolidation.

• Using intelligent caching: Mosaic's in-memory engine caches frequently run queries, reducing repeated hits to costly data warehouses and lowering compute strain.

• Offering predictable pricing: With a user-based license, you get predictable spend, avoiding the surprises of a compute-driven pricing model.

Saurabh Abhyankar, Chief Product Officer, Strategy:

"We charge per user—not based on query volume or compute. So if you have many expensive queries hitting your data warehouse, you can cache them in Mosaic. Once cached, they hit our engine instead of the database."

In an era where cost predictability is as important as capability, Mosaic gives organizations control over where data and analytics run, how they scale, and what they cost—without sacrificing governance or scale.

Problem #5: Environmental Complexity

Enterprises today span multiple clouds, tools, and teams. And that complexity is growing. From the 2025 global survey:

• 45.5% have enterprise-wide governance but only 16.2% use AI to proactively drive decisions.

• 39.6% operate in hybrid hub-and-spoke models, with both centralized governance and distributed analytics teams.

As organizations scale, complexity introduces fragility: duplicated models, conflicting metrics, redundant dashboards, and mismatched access rules.

PeggySue Werthessen, VP of Product, Strategy:

"The real problem isn't that you have multiple tools. It's that you have duplication of effort and inconsistency of information across those tools."

How Mosaic Solves Environmental Complexity

Rather than enforcing rigid control, Mosaic creates a shared governance framework that connects tools, clouds, and teams—without sacrificing flexibility.

Today, most organizations operate in federated models with a diverse mix of tools and data platforms. Complexity is inevitable. The real risk lies not in having diverse systems, but in having inconsistent logic, duplicated models, and disconnected governance across them.

PeggySue Werthessen, VP of Product, Strategy:

"Instead of rebuilding every time your tech stack shifts, Mosaic lets you reconfigure your models—like tiles—around new structures or tools."

Mosaic addresses this risk by acting as a data fabric—a single layer where business definitions, access rules, and modeling logic are created once and applied everywhere.

Mosaic ensures:

• All tools and users call the same definitions.

• Data access and logic are applied consistently.

• Changes ripple safely and predictably across the stack.

This ensures consistent data whether the endpoint is a dashboard, an AI agent, or a custom application.

PeggySue Werthessen, VP of Product, Strategy:

"That endpoint today may just be three or four BI tools. But in the future, it can be BI tools, AI agents, and applications you give your customers. We need to make sure they are coordinated."

Mosaic doesn't eliminate complexity, it orchestrates it. It allows each team, tool, or department to innovate within guardrails, ensuring alignment across a distributed analytics environment. As your data ecosystem evolves, Mosaic adapts with it, serving as the connective tissue that keeps intelligence trustworthy, explainable, and reusable at scale.