Data Quality Isn’t Just About Good Readings
If you work in industrial operations, you know your plant lives and dies by the accuracy of its data. Most plants have invested heavily in sensors, controls, and a PI System infrastructure to capture data from every corner of the operation.
But here’s the truth: even with the best infrastructure, bad data still happens.
We’re not just talking about broken sensors or missing values. Data quality issues are often hidden, subtle, and cumulative. A flatlined value here, a mistyped tag there, a refactored AF model no one documented, these seemingly small issues can quietly ripple downstream through your KPIs, calculations, reports, and AI models, distorting the truth about what’s happening in your plant.
To get ahead of these issues, industrial companies need both continuous monitoring and comprehensive data lineage. One without the other leaves dangerous blind spots.
What Is Monitoring? (And What It Catches)
Monitoring is the real-time surveillance layer that watches your data flows and asset health. In a PI System environment, that means keeping tabs on:
- Tag health — stale, flatlined, or dropped signals
- Bad value rates — how often data comes in bad or invalid
- Threshold violations — when values exceed safe or expected limits
- Scan rate mismatches — duplicate tags with different scan rates (yes, it happens, and it gets messy)
- Missing or late data — where partial or delayed records skew results
Why does this matter? Because monitoring is your early warning system. It catches problems early, ideally before they cause operational disruptions or end up embedded in your monthly KPI reports or AI models.
But monitoring alone can only tell you that something looks wrong right now. It can’t explain why, how long it’s been happening, or where else it matters.
What Is Data Lineage? (And What It Reveals)
Where monitoring is reactive, data lineage is explanatory. It’s the map of where your data came from, how it was transformed, and where it went next.
In the PI System world, data lineage means being able to trace:
- Which tags, templates, AF elements, and analyses fed this value
- Who made changes to a tag, element, or calculation and when
- Which displays, reports, or dashboards are using that value downstream
- What dependencies exist when you refactor or update a model
Why does this matter? Because without lineage:
- You don’t know if a value you’re monitoring is truly trustworthy
- You can’t safely troubleshoot or confidently deploy AI
- You can’t measure the operational or financial impact of a bad value
Lineage is your source-of-truth map. It’s how you understand the full context and downstream effects of every data issue.
Why You Need Both, Not One or the Other
Here’s the critical thing: monitoring and lineage solve different, complementary problems.
Monitoring alone:
- Detects bad values
- Alerts you when thresholds are exceeded or data stops flowing
- Can’t tell you where the issue originated, what else it’s affecting, or why it matters
Lineage alone:
- Shows you the data’s path and dependencies
- Provides the historical context and impact analysis
- Doesn’t catch real-time issues as they happen
Together, they give you industrial data quality control.
- Detect problems quickly
- Show exactly where they started and where they’ve spread
- Help you prioritize what to fix first
- Prevent cascading errors into reports, AI models, and business decisions
- Build operational trust in your data-driven systems
What This Looks Like in the PI System
Let’s say a flow meter tag flatlines for 4 hours.
Without monitoring:
You might not notice until a report shows suspiciously low throughput the next day.
Without lineage:
Even if you catch the bad value, you’ll waste hours manually hunting through displays, AF templates, totalizers, and reports trying to figure out what it affected.
With both:
- Monitoring instantly alerts you to the flatline
- Lineage shows:
- That the tag feeds 3 downstream KPIs
- It’s used in an AF totalizer
- A production report with that value was emailed to leadership this morning
- Now you know:
- Exactly what to fix
- Which reports need to be corrected
- What operational decisions might have been made based on bad data
You just saved hours of fire drills, bad decisions, and potentially millions in operational risk.
Data-Driven Industrials Can’t Afford Blind Spots
The stakes have never been higher. With the surge in AI projects, regulatory requirements, and operational risk pressures, the days of trusting unverified process data are over.
Data quality isn’t just a data team issue anymore, it’s a business resilience issue.
If you’re serious about using PI System data to run your plant better, improve efficiency, and fuel AI projects you can trust, you need to be equally serious about both monitoring and lineage.
They’re not optional. They’re foundational.
👌 Wrapping Up
Data monitoring keeps you ahead of problems.
Data lineage explains what happened and what matters.
Together, they protect your operations, AI projects, and bottom line.
If your industrial data systems don’t have both, you’re running blind.
Ready to See It in Action?
If your team depends on PI System data to keep your plant running safely and efficiently. Osprey is built for you.
Tycho Data 