Why Data Quality Alone Isn’t Enough: Prioritizing What Matters with Osprey

A large PI environment can contain thousands of data-quality findings. Fixing every finding immediately is neither practical nor necessary.

The important question is not only "Is this tag healthy?" It is also "What depends on this tag?"

Add operational context to each finding

A quality issue becomes easier to prioritize when the team knows:

• Asset and site

• Asset criticality

• PI Vision usage

• AF analysis dependencies

• External consumers

• Business owner

• Age of the issue

This context separates low-value cleanup from high-impact operational risk.

Use risk, not issue count

A site with 1,000 low-priority stale tags can have less risk than a site with one failed signal that feeds a critical calculation.

Do not use total issue count as the main performance metric.

Distinguish active problems from technical debt

Useful categories include:

• Active operational issue

• Data-quality issue with limited impact

• Cleanup candidate

• Accepted exception

• Retired or intentionally inactive

This prevents the backlog from becoming a permanent list of undifferentiated warnings.

Track dependencies before retirement

A tag can look obsolete but still support a display, analysis, or external report.

Usage and lineage should be part of the retirement decision.

Assign ownership

Prioritization has little value if no team owns the repair.

Route the finding to the group that can act on the source, PI configuration, AF model, calculation, or downstream application.

Measure time and recurrence

Track how long high-priority issues remain open and whether the same failure returns.

Repeated issues often indicate a governance or monitoring weakness rather than an isolated technical problem.

The objective

A mature data-quality program does not try to make every PI Point perfect.

It identifies which data supports important decisions, detects when that data is unreliable, and directs engineering time to the findings with the highest operational consequence.