From data-driven talk to data-washing reality

What comes to mind when you hear phrases like “being data-driven,” “using artificial intelligence,” or “making data-based decisions”? Sound familiar? Most likely. In Latin America, these expressions have become a routine part of corporate discourse. They appear in strategic plans, board presentations, and conversations about digital transformation.

According to a survey on the state of data quality in Latin America conducted by Merlin Data Quality, organizations across the region have steadily increased the importance they place on data in recent years. There is greater awareness of the opportunities that information systems can create. However, this does not necessarily translate into actual implementation. The same report shows that there is still significant work to be done in terms of formal policies, defined roles, data quality, and established practices.

This gap between stated ambition and the actual maturity of data practices creates an environment that may conduct to gradual growth, but also gives rise to an increasingly visible phenomenon: “data-washing.”

Rather than outright “deception,” it is more useful to frame the term as a gap: the gap between the narrative and the real capability of data management. Just as the AI-Washing Report 2025 by Signal AI describes the rhetorical use of artificial intelligence without the technical or organizational structures needed to support it, data-washing can be understood as the adoption of the language of “being data-driven” without yet having robust practices for data governance, quality, and interoperability in place. Data-washing emerges when data takes a central role in strategic storytelling but remains fragile as an operational asset.

As a result, the conversation around data begins to reveal a tension. Declaring a “data-driven” orientation or investing in new analytics tools is no longer enough. This phenomenon can emerge when “working with data” becomes primarily associated with visible outcomes: dashboards, real-time indicators, or automated reports, without a deep review of how that data is produced, integrated, and validated.

The scale of this gap is significant. Globally, nearly 70% of organizations say data management is a priority for their digital transformation efforts; however, only about 20% report having a fully implemented enterprise-wide strategy. In the region, the issue becomes even more pronounced: approximately 66% of Latin American organizations say poor data management exposes them to regulatory sanctions, while nearly 70% cite the shortage of skilled professionals as a major barrier. This tension between discourse and execution is, in essence, the quantitative definition of data-washing.

Data-First: building intelligent systems on reliable data

The data-first premise responds to a strategic principle. It is fundamentally about defining priorities. When a database is weak, incomplete, or lacks governance, intelligent systems are built differently. Inconsistencies are replicated, errors are amplified, and failure processes become opaque.

Artificial intelligence, in particular, does not “fix” data problems, it amplifies them. Models trained on inconsistent, incomplete, or poorly integrated information reproduce those flaws at a greater scale. A survey by PwC indicates that 79% of organizations are adopting AI agents in some form; in this context, data quality issues can generate unpredictable model behavior, including hallucinated outputs or gradual performance drift. And with AI spending already exceeding US$ 2 trillion in 2026, tolerance for error decreases dramatically: as investment in AI scales, the cost of poor-quality data scales with it.

But what does a truly data-driven strategy actually involve? It is not simply about accumulating information. Nor is it about centralizing data in a repository. It is about defining how data flows across the organization, from capture to decision-making.

At a minimum, this type of plan requires clear rules regarding information quality and consistency, integration mechanisms between systems, explicit accountability for data, and a defined architecture.

There is also an internal dimension that few organizations openly acknowledge: when employees themselves do not trust the company’s data, decisions are made based on intuition or politics rather than evidence. Accenture found that only one out of three executives trust their data enough to derive real business value from it. This silent distrust may be the most expensive form of data-washing because it never appears on a dashboard or in a digital transformation report.

Clear signs of data-washing within organizations

From a technical perspective, data-washing is expressed less through what organizations say and more through the decisions they postpone. One example is the presence of dashboards without data lineage, information that lacks traceability or clear visibility into its origin and transformation process.

Another common sign is the passive centralization of information: centralized repositories where data is merely accumulated without consistent normalization processes or clear reuse criteria. Rather than enabling scale, this approach often turns “centralized data” into a bottleneck that is difficult to audit, maintain, or evolve.

Added to this are the so-called “quick fixes” that solve isolated business problems but, instead of integrating into the broader ecosystem, deepen fragmentation across the overall architecture.

However, the clearest indicator appears in the use of artificial intelligence. This happens when projects operate on manually prepared datasets disconnected from core systems, or when, after an error occurs, no one can clearly explain who owns the data or under which rules it was generated.

But data-washing also has cultural and organizational roots that go beyond technical limitations. In many Latin American organizations, data represents political power: sharing information across departments is perceived as a threat, and nobody wants to “own” a dataset because ownership implies accountability. This dynamic creates silos that are not only technological but relational. The result is a fragmented data architecture that ultimately reflects an organizational structure never designed for collaboration.

This is further intensified by growing regulatory pressure. Latin America is advancing toward stronger data protection frameworks, with Brazil’s LGPD serving as the region’s most consolidated reference point, while the region progressively adopts standards similar to Europe’s GDPR. This introduces a legal-risk dimension that makes data-washing potentially very costly: the issue is no longer limited to fragile decision-making but extends to concrete regulatory exposure. IBM Institute for Business Value estimated in 2025 that more than a quarter of organizations lose over US$ 5 million annually due to data quality problems, with 7% reporting losses exceeding US$ 25 million.

Cases that illustrate the opposite path

Against these signs of data-washing, there are also organizations across the region that have taken the opposite approach: they did not begin with the promise of artificial intelligence or flashy visualizations, but by first addressing their data foundation as infrastructure.

The goal? Integrate business data, standardize processes, and define interoperability standards. In consumer industries such as Hypera Pharma, the emphasis has been placed on consolidating operational, financial, and logistics data before introducing any layer of intelligent technology. Likewise, in the financial sector, Santander prioritized data traceability and consistency due to the sector’s regulatory and omnichannel complexity. At Arauco, the consolidation of forestry operations, industrial plant, and financial data into an integrated architecture preceded any analytics initiative, with a strong focus on standardization and traceability.

Because working with data means recognizing the need to build infrastructure where data can move, feed back into systems, and generate value. And like any infrastructure, if it is not properly designed and maintained, it creates not value but the opposite: friction, noise, and poor decisions. The challenge for organizations in Latin America is to become data-driven without falling into data-washing. And that challenge is technical, regulatory, and above all, cultural.