What data leaders need to know about modern data quality

Data Quality and Data Governance are important, however, nowadays, “important” just doesn’t cut it anymore. There is a saying that perfectly reflects modern reality:

“Adequate is no longer acceptable, and acceptable is no longer adequate.”

Data Quality and Data Governance are essential for building successful, high-performing data platforms. If you're not doing both, you're missing out.

Why data quality matters

Organizations that have undertaken data-driven initiatives often encounter challenges where poor data quality delays outcomes. Gartner estimated that poor data quality costs companies on average $12.9 million and significantly complicates decision-making.

It starts with operational systems like CRM, where poor form design or flawed processes lead to duplicates, inaccuracies, or other data issues. Lack of observability can lead to campaign misfires or poor customer experiences. For example, a coding mistake during a migration project led Equifax to send incorrect credit scores to millions of customers - something that could’ve been caught early with proper data observability. (I have a particular rule in mind that Collibra Data Quality provides OOTB!).

Data analysts frequently encounter inconsistent territory codes across tools like spreadsheets, Tableau, Power BI, or Salesforce. You can find all kinds of stuff in there, country, state, city or free-form text that breaks calculations, requiring additional remediation or clarifications with business teams. And that is just the simplest example.  

This highlights the importance of a robust Data Quality framework. 

It spans both operational and analytical systems, but what makes it stand out, you guessed it, is AI. The numbers for delayed or failed AI projects are quite staggering, from a conservative 30% to   80%+ with one of the significant reasons for failure being data quality issues.

CIO Priorities 2025 priorities report by Info Tech research group confirms the same, that Data Quality is the top challenge and obstacle for CIOs implementing AI. Data Quality suffers the highest, having the lowest satisfaction score in that report.  

But where do we start? How do we know where and which data sets to observe if we don’t understand our data first? Theoretically, even if you can monitor every data set in your company, considering cloud consumption models, it is going to be a pretty penny. That’s where data governance plays its role.

Long live Data Governance!

Let me share a quick anecdote. Back in 2007, I was a back-end engineer at a Telco company, occasionally handling ad-hoc reports. One day, “business” asked for a report on new customers by product for the past quarter.

I thought it would be easy. I knew the system well. I pulled the schema, built and tested the query, got it into SIT, and shared the report.

It was wrong. And so were the next few versions. Only after a call did we realize we had different definitions of "customer." I tried purchases, then active subscriptions, but the right one turned out to be: records where the business party number (BPN) is active.

Still, the numbers didn’t match. After nearly two months, we traced the issue to BPN duplication due to inconsistent formats across sales channels. We fixed the report and kicked off a small project to address the data quality issue.

That was the norm back then.

Looking back, it was a classic case of time wasted, something a modern data governance program easily resolves today. The foundation of common understanding of terminology through terminology through an enterprise glossary, business lineage down to the system of record, and data quality monitoring of critical data elements such as business party number.

Any confusion today can be resolved in minutes by tagging the right data or business teams directly in your data governance platform.

Depending on your data maturity, you might go a step further - using a report catalog to find existing reports instead of recreating them, and simply request access through data shopping.

To improve Data Quality, organizations should begin by establishing a strong Data Governance foundation. A strong foundation includes:  

• Data quality policies and standards
• Clear terminology through business glossary
• Catalog of critical data elements
• Good communication channel between data consumers and data teams
• Ownership and accountability of the data team and domain owners
• Issue management and resolution process
• Transparency and visibility through business lineage

Check out Gartner’s article on the above recommendations, it is a good way to start improving your data quality.

More than that, according to Gartner, a robust data governance framework will fail to recognize value from their AI project. So, with a strong data governance foundation, you are contributing to both – your data quality and mitigating risks in your AI projects.

However, the existing organizational structure and processes alone is not enough. As seen in the Equifax example, our goal is to establish observability that empowers both business and data teams to take control of our data. That’s no longer possible with legacy processes or outdated software.

Why the platform matters

We can observe a current trend toward shifting data quality controls earlier in the data ingestion process, often referred to as "Data Quality shift left." We see a “shift left” happening across the industry to improve the quality of deliverables and boost communication in the industry. This approach emphasizes conducting data quality checks and controls closer to the source rather than monitoring assets nearer to the business. While this initiative is beneficial, it carries certain risks.

In many organizations, IT teams often build new data platforms without integrating Data Quality in the pilot phase, believing that all necessary controls could be implemented within the Data Platform by the IT team. They proposed that pipeline metrics would be shared with the Data Team, which they thought would establish "proactive" data quality. Similar oversights have occurred in well-documented data incidents, such as the Equifax breach. 

Yes, testing data in the data pipeline is a must in modern data engineering practices and platforms. However, relying solely on pipeline monitoring is not enough. Nowadays, data is a dynamic area. Apart from the growing number of different regulations and acts coming into the force, increasing number of vendors and platforms – business doesn’t know what they don’t know. Like in the Equifax scenario, the right tool would’ve at least allowed the business owner to establish observability over their data domain. Or even better, the right Data Quality tool would automatically detect an anomaly or inconsistent pattern and alert the data owner. The "shift left" approach alone cannot achieve a unified observability strategy. The data team should not be a passive receiver of metrics; they must actively participate in ensuring Data Quality. As part of a comprehensive observability strategy, it is essential for your data team to utilize a robust data quality platform.

But as we said earlier, adequate is no longer acceptable. Manually discovering data quality issues and running mini-projects to fix them, like it’s still 2007, only slows down your business and innovation.

For traditional platforms, let’s use SAP Information Steward (SAP IS) as an example, while effective for their time, were built in an era before cloud-native architecture, AI observability, and real-time responsiveness became essential. They often rely heavily on static rules and batch-based analysis, limiting agility and visibility in today’s data ecosystems. In contrast, modern solutions like Collibra Data Quality are designed for scale, learning, and automation—offering a paradigm shift in how data issues are identified, tracked, and resolved. It is the way to step up your data game.

Why Collibra DQ is a modern alternative to legacy tools like SAP IS

Reactive vs. Proactive

Legacy platforms like SAP IS depends on known issues and pre-built controls. But what about unknown issues? Collibra DQ uses behavior-based learning to detect anomalies proactively, often before users realize there's a problem.

As a result with legacy platforms, like SAP IS, data issues may linger until manually investigated. Collibra DQ prevents data contamination with automated early detection and alerts.

Static Rule-Based Approach vs. AI/ML-Driven Monitoring

SAP IS relies on fixed rule sets. Collibra DQ evolves. It uses AI/ML to detect new and changing patterns in your data. It lets your data quality steward teach Collibra AI about your company’s data, helping it learn what’s a bug and what’s a feature.

Data stewards can still configure use case specific rules—using an AI Copilot that makes it easy even for non-technical users—ensuring flexibility without dependency on niche skillsets.

Batch Processing vs. Real-Time Monitoring

SAP IS is optimized for batch processing, which delays issue resolution. Collibra DQ supports near-real-time monitoring and offers out-of-the-box source monitors and schema drift detection. This ensures immediate insights and faster action.

Isolated Governance vs. Unified Governance

Data Governance is foundational to effective Data Quality. That’s why your Data Quality platform must integrate seamlessly with your enterprise governance system. Collibra excels at providing unified observability, which allows you to be proactive with data quality issues, identify gaps, take ownership, and hold accountability. This integration can significantly reduce the time to market (TTM) for your data products and decrease the time needed for issue resolution.

Why organizations are moving away from legacy DQ platforms

We’ve worked with several customers who have transitioned from legacy DQ platforms and processes to Collibra DQ to meet modern data governance and quality demands. Key reasons for the switch include ease of use, integration with Collibra Data Intelligence Platform, AI-powered monitoring and governance, faster onboarding for data teams and ability to work in hybrid environment with modern data platforms like Deatricks or Snowflake.

With strong governance and data quality frameworks, our customers have seen up to 180% ROI in the first year. Plus, big improvements in analytics (40%+ accuracy boosts and better communication) and data processes (shorter TTM and over 50% faster data access).

How do we get started   

Your success depends on your organizational maturity. Establishing a solid Data Governance foundation is crucial, as it paves the way for effective Data Quality and brings immense benefits to your organization.

In addition to Gartner's approach to improving Data Quality, here are some key steps based on our experience:

• Find your business partner: Target a domain with known data quality issues.
• Set clear goals: Define measurable benefits.
• Start small, think big: Keep early wins tight and focused.
• Show real results: Make them tangible for end users—not just in slides.

FWD view team (now a part of Nagarro) has been a long-time Collibra partner, offers proven accelerators to help you onboard CDQ faster. Our 4–8-week programs include:

• Best practices setup
• Automation software
• End-to-end implementation

A few examples at the Collibra marketplace include

• Data Quality Accelerator by FWD view (now Nagarro) at the Collibra marketplace.
• Masking Accelerator

If you have any questions,  please reach out to your customer success manager at Collibra or to Jevgenijs.jelistratovs@nagarro.com . We are always happy to discuss Collibra and Data Governance.  

Don’t miss out, enhance your Data Quality with Collibra DQ.