Detect and Respond to Data Quality Changes

Learning Objectives

After completing this unit, you’ll be able to:

• Explain how organizations detect meaningful changes in data quality over time.
  
• Describe how profiling snapshots, dashboards, automation, and stakeholder collaboration support continuous monitoring.
  
• Differentiate between expected business changes and data quality issues that require corrective action.
  

Follow the Data Quality Lifecycle

Data quality isn’t static. Business processes evolve, new fields are introduced, users adopt new ways of working, and integrations continuously change both data and metadata. As mentioned at the beginning of this badge, it’s common to move forward in the process, then revisit previous activities as things change. And we do something similar in this unit—revisit concepts covered in previous units with additional context.

Manage a Continuous Process

Effective monitoring follows a repeatable process.

• Profile data on a scheduled cadence based on business needs.
  
• Compare current results with previous profiling snapshots to detect meaningful deviations.
  
• Investigate the cause to determine whether the change is expected or requires action.
  
• Respond by improving data, business processes, or metadata configuration.
  
• Continue monitoring to confirm the expected outcome.
  

Not every detected change is a problem, however.

NTO Example: From Monitoring Opportunity Data

Luna schedules monthly profiling of opportunity data to monitor how sales processes and field adoption evolve over time. Rather than beginning with individual fields, Luna first compares the latest profiling snapshot with previous profiling results to understand whether the overall data capture patterns across the opportunity object have changed.

Reviewing the latest profiling snapshot, Luna notices several changes.

• More fields now fall within the 30–50% fill-rate range.
  
• Fewer fields remain within the 50–75% range.
  
• Slightly more fields have become 75–100% populated.
  
• The number of fields that are always populated (100%) has remained stable.
  

These changes suggest that data capture patterns have shifted across multiple fields rather than a single field becoming unreliable.

Instead of assuming a data quality issue, Luna drills into the individual profiling results to determine which fields moved between categories and why.

Working with sales operations, she discovers that a recent opportunity process update introduced several optional fields while moving some existing information later in the sales lifecycle. As a result, some fields became more widely adopted, while others were intentionally used less frequently because the information is now captured elsewhere in the process.

Because the profiling results match the planned business process change, Luna determines that no corrective action is required. Instead, she updates the monitoring baseline so future profiling runs compare against the new expected data capture patterns.

If the investigation had instead revealed unexpected declines caused by low user adoption, integration failures, or configuration issues, Luna could have coordinated with business stakeholders to implement corrective actions and validated the improvements during the next scheduled profiling cycle.

Monitoring Methods Work Together

Different monitoring methods answer different questions. Let’s revisit the monitoring methodology from the Monitoring Methods and Cadence unit, focusing on the types of questions each method is essentially answering.

Together, these monitoring methods provide a complete picture of data quality and help organizations respond before issues impact users, business processes, analytics, automation, or AI. In the above example, Luna used a dashboard and scheduled profiling to see what the data looked like today compared to the past, ultimately getting at what changed over time. She got stakeholder feedback which told her that the old snapshot of data did not support the business process anymore. So, she updated the baseline.

Module Recap

Monitoring completes the data quality lifecycle. By combining scheduled profiling, business readiness KPIs, dashboards, alerts, and stakeholder feedback, organizations can detect meaningful changes, investigate their causes, distinguish expected business evolution from data quality issues, and validate that improvements are effective.

Throughout this series, you’ve learned how to assess, improve, enrich, unify, and continuously monitor data quality. Together, these practices help organizations maintain trusted data that supports people, processes, analytics, automation, and AI as business needs evolve.

Resources

• Trailhead: Explore Data Quality Fundamentals
• Trailhead: Prepare Your Data for AI