“Many CDAOs fall into the trap of data hoarding or paranoid data control. Data itself is an example of entropy and a better model of balancing controlled chaos with observable behavior helps reduce the friction between governance, application developers and users.”

“As a CDAO, it is important to avoid falling into the trap of assuming that one data design is “better” than another. The moment a system or application is imbued with authority—instead of the data content—the use of the data from that application becomes limited.”

“Instead of planning to integrate data “later”, start trying to reuse it immediately upon deployment. Compare it to other similar data right away—and where possible automate that process.”

Gartner recommends that CDAOs focus on seven principles of data management:

• Learn from users already accessing data.

• Don’t design from scratch or assume you will redeploy data.

• Find as much metadata as you can.

• Purposeful redundancy is necessary.

• Start with passive metadata, but know that you will need active metadata capabilities.

• Existing data management solutions work. Focus on them working together. 

• Integration, data quality, data mastering and security do not go away.

 A Gartner survey found that 46% of AI projects never make it into production. “The problem is that many AI projects hit a wall. We often make some good initial progress in AI projects, but as time goes by we hit a performance barrier.”

“When we ask organizations what are their key barriers for AI implementation, measuring and understanding business value comes right at the top.”

Don’t confuse model performance with business performance. Models make predictions, but capturing business value requires you to act. We need to redesign our process and measure the factors that actually drive business value.”

Gartner developed an AI compass focused on four cardinal directions - Strategy, Data, Modeling and Testing and Infrastructure and Operationalization - to help teams navigate the wide range of tactics to improve performance.

Five key actions to take to push stalled AI projects into production:

• Share the AI Compass with teams and use it to prioritize next actions

• Involved domain experts in AI projects

• Expand data collection internally and externally

• Define an action plan for AI projects to capture value

• Invest in AI engineering to accelerate and improve AI value at scale

“Data observability provides continuous, holistic, and cross-sectional visibility into complex data landscapes and synthesizes signals across infrastructure, application, and data layers to provide a comprehensive understanding of individual components, data pipelines, and system performance.”

“It allows the enterprise to measure what matters in their data journey, experiment and adapt, and implement data strategies that align with business strategies and requirements.”

Data observability primarily focuses on five things: Observing data; observing data pipeline; observing data infrastructure; observing data users; and observing cost and financial impacts.

“Data observability is powerful, but has limits, such as only covering data at rest, not data in motion or doesn’t deliver data or fix data issues.”

“Partner with business stakeholders to evaluate and demonstrate the business value of data observability practices by tracking improvement of data issues management within data pipelines to show tangible benefits.”