Lack of data ownership leads to failed AI implementations

Data has always been at the heart of business decision making. The quality of data, therefore, has an immense impact on business results. And now the rapid emergence of Gen AI has placed an even greater importance on the quality of data, and more specifically its ownership and management within an organization. But a failure to manage your data will lead to failed AI implementation, before its even started!

This dramatic rise of Gen AI-based systems is fuelling the demand for high-quality, bias-free, reliable data. In our latest World Quality Report, the 16^th edition of the industry’s largest research study looking at the current state of Quality Engineering practices around the world, we focus on ‘Data Quality’ and its impact on usage within Gen AI-based solutions.

We had expected to see an increased focus on data in this year’s Report, with organizations placing greater emphasis on its importance as part of their Gen AI innovations. But what we found was disappointing.

Assessing data’s impact: The current landscape

This year’s survey results reveal that 16% of organizations consider data critical, while a further 48% see it as very important. Conversely, only 1% do not regard it as important at all. What this tells us is that while 64% of organizations acknowledge the significance of data, 36% still do not prioritize it as they should.

These findings are unwelcome, especially given the severe consequences of data breaches and the fundamental role data plays in organizational operations and AI systems. Add to this the stringent GDPR legislation and the European AI Act. So, this lack of prioritization raises concerns about organizations’ operational efficiency, decision-making, compliance, and potential reputational damage.

The value in data management

When we drill down further to try to understand the value organizations are placing on their data and its management, we see that despite a third of organizations not viewing data as crucial, its value in enhancing customer satisfaction and reducing costs is clearly evident:

• Increased production quality (56%) – Improved data management enhances production quality, reducing errors while boosting efficiency in business operations.
• Better customer outcomes (49%) – High-quality data leads to more personalized and effective customer interactions, increasing satisfaction and loyalty.
• Cost reduction (45%) – Effective data management reduces operational costs by reducing inefficiencies and errors.

Measures to keep test data accurate and free from bias

While many organizations recognize the importance of regularly updating or refreshing their test data (44%), we see that more and more organizations are turning to AI-generated test data (49%) instead. This is a concerning trend as it comes with its own set of ethical, security, and compliance issues, as AI-test data is highly likely to have bias embedded.

Almost a third of organizations rely on AI to check data quality and remove biases (34%), but this approach often lacks transparency and context, which can unintentionally reinforce existing biases. While synthetic data generation tools are gaining traction (49%), it is crucial to maintain data integrity and ensure compliance.

It therefore becomes essential to address these issues in order to maintain data integrity, develop reliable AI systems, and navigate the complex ethical and legal landscape. In order to achieve this, our key recommendations are:

1. Establish clear ownership and responsibility for management and quality of all relevant data across the organization. This includes business related data, as well as your critical technical support data.
2. Invest in high quality, bias-free data through AI tools to support better customer outcomes, and informed decision-making.
3. Embrace legal frameworks like GDPR, to ensure compliance and protect data integrity.
4. Recognize data’s value and the critical role data plays in their smooth operations. Adopting this mindset will enable better efficiency and decision-making.

In summary…

For over 15 years, we’ve been asking questions about data and its importance in the World Quality Report. Each year, organizations talk about focusing more on this, yet the same perceptions persist about the quality and importance of data. Despite the critical role data plays in AI and organizational success, many organizations still do not give it the focus it deserves.

From an ethical standpoint, we as IT professionals must ensure that data is accurate, fair, and as free from bias as possible. Arguments about the costs of tools or the lack of appropriate tools have been ongoing for years. With the demands of AI, compliance requirements for data access, the need for effective testing in cloud environments, and the associated costs, these issues must be managed now.

To find out more, download a copy of the 2025 World Quality Report.